RiskRADAR - CICTT Taxonomy Review

[PROBABLE CAUSE] I to be accomplished on through flights or at trip termination whenever time is less than 12 hours per Maintenance Manual Procedures BX 12-0-1-1. Aircraft with layover of 12 hours or more will receive a Service No. 2 not to exceed 65 flight hours between checks. 108 APPENDIX D J *"NJURY INFORMATION . .--The second officer sustained minor superficial brush bums to both elbows and forearms, during the evacuation. Cabin Crewmembers.--The cabin crewmembers sustained the following injuries during the evacuation: Flight attendant No. 1 sustained a strained left shoulder; Flight attendant No. 2 sustained acute thoracic and lumbosacral strain; Flight attendant No. 3 sustained a mild right bicep strain; Flight attendant No. 4 sustained a left elbow contusion, left shoulder dislocation, and mild lumbosacral strain; Flight attendant No. 5 sustained a left calf contusion; Flight attendant No. 6 sustained a mild left elbow bruise; Flight attendant No. 7 sustained mild left arm and lower back strain; Flight attendant No. 8 sustained a soft tissue injury to the back; Flight attendant No. 9 sustained abrasions to both palms and the left knee; Flight attendant No. 10 sustained a fracture of the left tenth rib; Flight attendant No. 11 sustained a minimal injury to the right middle finger PIP joint and left first MP joint; Flight attendant No. 12 sustained a pulled muscle on the left side of the neck; 109 Flight attendant No. 13 sustained a comminuted fracture of the e right ulna and radius; , Flight attendant No. 14 sustained a mild thoracic back strain; Flight attendant No. 15 sustained a non-displaced fracture of C-6, a cerebral concussion, a fracture of the proximai right humerus, and multiple lacerations; A flight attendant, flying as a passenger, sustained mild lumbosacral strain, a laceration of the right little finger, and a left elbov abrasion. Passengers.--Nine Passengers who were seated in seats 8)1, 9FGH, IOGH, 11GH, and 12H, were ejected from the fuselage and were not found; and thus, are assumed to have been fatally injured in the accident.

[PROBABLE CAUSE] Passengers.--Nine Passengers who were seated in seats 8)1, 9FGH, IOGH, 11GH, and 12H, were ejected from the fuselage and were not found; and thus, are assumed to have been fatally injured in the accident. Passengers seated in the indicated seats sustained the following injuries: peat 1c Barotrauma to both ears 9C_ - Half-inch laceration to the upper left arm, superficial abrasions to left arm and hand, barotrauma to both ears 9E - Superficial abrasions and contusions to the left hand, mild barotrauma to both ears 10B - Superficial abrasions to the left elbow and left middle finger 1OE - Superficial abrasions to the torso and left forearm, bruising of the left hand and fingers ME - Laceration on the right ankle tendon, multiple bruises LIF 13E 13H i4A 15J 16B 16) 16K 26A 26B 26H 27A 28) 110 Slight contusion of the right shoulder Barotrauma to both ears Bleeding in ooth ears Contusion to the left periorbital area Lareration in the parietal occipital area, barotrauma to both ears Comminuted fracture of the lateral epicondyle of the left distal humerus (about 5mm separation) Superficial abrasions to the right arm Barotrauma to both ears Right temporal abrasions Barotrauma to both ears barotrauma to both ears Barotitis to both ears, low back pain, irritation to the right eye due to foreign bodies Barotrauma to the right ear Superficial abrasions and a contusion to the left hand, mild barotrauma to both ears MT ON Rt nee are HNL HNL AXL SY0 AKL KN KNL AX +) £9) AX REE IR80UND FLT/DATE 830 12/5 824 32 git 811 611 832 82 812 8h) 811 811 812 812 $12 811 $13 12/6 12/7 12/7 4/7 12/7 12/9 12/9 12/9 12/9 12/9 12/9 12/1}} 12/11 12/12 12/32 12/12 Ii APPENDIX E MAINTENANCE HISTORY OF N4713U OUTBOUND FLT/OATE 825 bi2 811 $11 811 812 812 Sle 811 611 811 i2 Bi2 si2 613 813 $11 12/5 12/7 42/7 12/7 12/7 12/9 12/9 12/9 12/9 12/9 12/9 12/11 12/11 12/12 12/32 32/12 12/12 PROBLEM Repcert = forward sarge door wil}} rot open.

[CONCLUSIONS > FINDINGS] The Federal Aviation Administration’s (FAA) current recommended brace positions do not take into account newly designed seats that do not have a breakover feature, and, in this accident, the FAA-recommended brace position might have contributed to the shoulder fractures of two passengers. 35. The flight attendants initiated the evacuation promptly, and, although they all encountered difficulties at their exits, they still managed an effective and timely evacuation. 36. Although the airplane was not required by Federal Aviation Administration regulations to be equipped for extended overwater operations to conduct the accident flight, the fact that the airplane was so equipped, including the availability of the forward slide/rafts, contributed to the lack of fatalities and the low number of serious cold-water immersion-related injuries because about 64 occupants used the forward slide/rafts after the ditching. 37. The determination of cabin safety equipment locations on the A320 airplane did not consider that the probable structural damage and leakage sustained during a ditching would include significant aft fuselage breaching and subsequent water entry into the aft area of the airplane, which prevents the aft slide/rafts from being available for use during an evacuation. 38. Given the circumstances of this accident and the large number of airports located near water and of flights flown over water, passenger immersion protection needs to be considered for nonextended-overwater (EOW) operations, as well as EOW operations. 39. If the life lines had been retrieved, they could have been used to assist passengers on both wings, possibly preventing passengers from falling into the water. 40. Equipping aircraft with flotation seat cushions and life vests on all flights, regardless of the route, will provide passengers the benefits of water buoyancy and stability in the event of an accident involving water. 41. Briefing passengers on, and demonstrating the use of, all flotation equipment installed on an airplane on all flights, regardless of the route, will improve the chances that the equipment will be effectively used during an accident involving water. 42. Passenger behavior on the accident flight indicated that most passengers will not wait 7 to 8 seconds, the reported average life vest retrieval time, before abandoning the retrieval attempt and evacuating without a life vest. 122 NTSB Aircraft Accident Report 43. The current life vest design standards contained in Technical Standard Order-C13f do not ensure that passengers can quickly or correctly don life vests. 44. Most of the passengers did not pay attention to the oral preflight safety briefing or read the safety information card before the accident flight, indicating that more creative and effective methods of conveying safety information to passengers are needed because of the risks associated with passengers not being aware of safety equipment. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the ingestion of large birds into each engine, which resulted in an almost total loss of thrust in both engines and the subsequent ditching on the Hudson River.

[ANALYSIS] NTSB Aircraft Accident Report 106 inflation handle, postaccident video analysis indicated that about 20 seconds elapsed after she opened door 1L until the slide/raft began to inflate. Further, both of the first two passengers to reach door 1L reported that the flight attendant seemed to be struggling with the slide/raft. The delay prompted one passenger to jump into the water before the slide/raft was inflated. Examination of the slide/raft at door 1L did not reveal any evidence as to why the slide/raft did not inflate automatically. Flight attendant C reported no difficulties opening door 1R; however, she stated that, sometime after the 1R slide/raft finished automatically inflating, she noticed that the door had closed somewhat and was impinging on the slide/raft. Although the slide/raft was preventing the door from closing any further, she had a passenger help keep the door as far open as possible so that it would not puncture the slide/raft. Postaccident recovery of the airplane severely twisted door 1R. A visual examination of the 1R door gust lock mechanism was inconclusive as to the cause of the anomaly. Passengers and flight attendants reported that the evacuation was relatively orderly and timely. A review of passenger exit usage indicated that, in general, passengers from the forward and mid parts of the cabin evacuated through the exit closest to their seats. (See figure 9.) However, aft-seated passengers indicated that water immediately entered the aft area of the airplane after impact and that the water rose to the level of their seat pans within seconds; therefore, they were not able to exit from their closest exits because these exits were no longer usable. Most of the aft passengers initially attempted to go to the rear exits, but flight attendant B and several passengers began shouting for everyone to go forward because the rear exits were not usable. The passengers turned around and attempted to go forward up the aisle, but, because a line had formed for the overwing exits, they initially could not move forward. The water in the back of the airplane rose quickly, which, in addition to improvised commands from flight attendant B to “go over the seats,” resulted in numerous passengers climbing forward over the seatbacks to reach a usable exit.142 However, some aft passengers remained in the aisle queue to the overwing exits. Many of these passengers noted that, when they arrived at the exits, the wings were crowded and people were exiting slowly. They also reported that the aisle forward of the overwing exits was completely clear and that the flight attendants were calling for passengers to come forward to the slide/rafts. Many of the aft passengers complied with these instructions; therefore, the 1L and 1R slide/rafts mostly contained passengers from the very forward and very aft parts of the cabin. The NTSB concludes that the flight attendants initiated the evacuation promptly and, that, although they all encountered difficulties at their exits, they still managed an effective and timely evacuation. 142 Not all of the aft passengers reported hearing the flight attendant’s improvised commands.

[ANALYSIS] Many of the aft passengers complied with these instructions; therefore, the 1L and 1R slide/rafts mostly contained passengers from the very forward and very aft parts of the cabin. The NTSB concludes that the flight attendants initiated the evacuation promptly and, that, although they all encountered difficulties at their exits, they still managed an effective and timely evacuation. 142 Not all of the aft passengers reported hearing the flight attendant’s improvised commands. NTSB Aircraft Accident Report 2.10.3 EOW and Ditching-Related Equipment 2.10.3.1 Slide/Rafts 2.10.3.1.1 Availability After a Ditching The accident airplane was equipped for EOW operations; however, the flight route from LGA to CLT was not an EOW route. Therefore, the flight could have been operated with a non-EOW-equipped airplane. The amount and type of safety equipment carried by EOWequipped airplanes differs greatly from that carried by non-EOW-equipped airplanes. Most significantly, EOW-equipped airplanes must carry passenger life vests and sufficient slide/rafts and/or life rafts to contain all of the airplane’s occupants even if one slide/raft or life raft of the largest capacity is unavailable. In contrast, non-EOW-equipped airplanes may operate with just evacuation slides and flotation seat cushions. (After the ditching, two slide/rafts on the accident airplane were unavailable because of water entry in the aft cabin.) The accident airplane was equipped with 4 slide/rafts, 2 at the front of the airplane and 2 at the back of the airplane, each of which was rated for 44 passengers with an overload capacity of 55 passengers. Because the two aft slide/rafts were unusable after water entered the airplane, only two rafts, with a combined capacity to carry 110 people, were available. However, given that this was a non-EOW flight, it was fortunate that the airplane was EOW equipped and, therefore, had any slide/rafts available at all for passenger use. According to information gathered from 146 of the passengers and the flight and cabin crewmembers, about 64 occupants were rescued from the forward slide/rafts, and about 87 occupants were rescued from the wings and off-wing ramp/slides, which were neither detachable nor considered part of the airplane’s EOW emergency equipment. Both passenger statements and photographic evidence, as shown in figure 2, indicated that the wings were very near to, if not at, standing capacity. Therefore, the wings did not have room for the additional 64 occupants who were rescued from the slide/rafts. If the airplane had not been EOW equipped, the rafts that held those occupants would not have been available. Further, at the public hearing, a US Airways representative stated that, if the accident airplane had not been equipped with slide/rafts, the flight attendants would have detached the single-lane slides at the forward doors and instructed passengers to jump into the water and hold onto them, exposing many passengers to cold water for sufficient time to likely cause serious injuries and/or fatalities.

[ANALYSIS] The NTSB notes that, after exiting the airplane through the overwing exits, at least nine passengers unintentionally fell into the water from the wings. 110 NTSB Aircraft Accident Report Therefore, the NTSB concludes that, if the life lines had been retrieved, they could have been used to assist passengers on both wings, possibly preventing passengers from falling into the water. The NTSB recommends that the FAA require 14 CFR Part 121, Part 135, and Part 91 Subpart K operators to provide information about life lines, if the airplane is equipped with them, to passengers to ensure that the life lines can be quickly and effectively retrieved and used. 2.10.3.3 Life Vests and Flotation Seat Cushions 2.10.3.3.1 Equipage Because the accident airplane was equipped for EOW operations, it carried life vests for both passengers and crewmembers. However, given that the accident flight route was not an EOW operation, the airplane could very well have been equipped with only slides and flotation seat cushions as the primary means for passenger flotation. In that case, flight attendants would have detached the forward slides and instructed passengers to jump into the water with their flotation seat cushions and hold onto the slide. If no slide/rafts had been available at the forward door exits, many of the passengers egressing from these exits would have had no choice but to jump into the water with no flotation device. (About 47 percent of the passengers did not exit with a flotation seat cushion.) Even if they had retrieved their flotation seat cushions, many passengers would have experienced extreme difficulty holding onto a seat cushion for more than a few minutes because of the effects of cold-water immersion. Self-righting life vests designed in accordance with TSO-C13f, such as those on the accident airplane, are designed to keep an individual’s head above water even after he or she is unable to swim or effectively move his or her arms and legs. In Safety Study 85/02, the NTSB issued Safety Recommendations A-85-35 through -37, which recommended that all Part 121, 125, and 135 passenger-carrying air carrier aircraft be equipped with approved life vests meeting the latest TSO and to ensure Part 25 requirements were consistent with the amendments made to Parts 121, 125, and 135. In response to these recommendations, on June 30, 1988, the FAA published NPRM 88-11, which proposed new requirements that “would ensure that each occupant is provided a life preserver which provides the basic benefits of high buoyancy and water stability…regardless of whether the airplane is involved in overwater operation.” In 1997, the FAA informed the NTSB that a final rule was expected to be published in the Federal Register by the end of that year; however, no final rule was issued. Subsequently, the FAA stated that “due to the amount of comments received and the amount of time since the NPRM was originally issued,” it had decided to publish an SNPRM by October 2000.

[ANALYSIS] The passengers who jumped or fell into the water (and the passengers on the wings who would have had to eventually enter the water if the emergency response had not been so timely) were at the most risk. Medical literature indicates that cold-water immersion causes cold shock, which can kill a person within 3 to 5 minutes, and swimming failure, which can kill a person within 5 to 30 minutes. Therefore, if the rescue vessels had not been near the accident site, it is likely that some of the airplane occupants would have drowned due to cold shock or swimming failure. 2.10 Survival Factors Issues 2.10.1 Accident-Related Injuries 2.10.1.1 FR65 Vertical Beam Flight attendant B sustained a deep, V-shaped laceration to her left shin during the accident. Although she could not remember being injured and only noticed the injury after she had evacuated the airplane, the investigation determined that the FR65 vertical beam had penetrated the floor directly beneath the aft, direct-view jumpseat on which flight attendant B had been seated. The shape of the beam matched the description and location of flight attendant 103 NTSB Aircraft Accident Report B’s injury. It is likely that she did not immediately notice the injury because of the shock of the impact and immediate submersion of her legs in near-freezing water. According to Airbus, the FR65 vertical beam is a nonstructural beam installed between the passenger and cargo floors at the aircraft centerline that is held in place by two quick-release, removable pins at its uppermost attachment point with the subfloor structure. Removing the pins and rotating the beam down allows maintenance personnel to access the waste water tank. Physical evidence indicated that, during the impact, the beam was pushed upward and rotated, allowing the removable pins to slide from the upper bracket and the beam to puncture the cabin floor above. In April 2009, an A321 was involved in a tail strike and incurred similar damage to the FR65 vertical beam; however, the beam did not puncture the floor. Airbus’ analysis of this incident and the accident event indicated that the damage to the accident airplane was more severe because of the continuous pressure applied to the fuselage skin by the water, which led to more skin and vertical beam movement. The NTSB concludes that flight attendant B was injured by the FR65 vertical beam after it punctured the cabin floor during impact and that, because of the beam’s location directly beneath the flight attendant’s aft, direct-view jumpseat, any individual seated in this location during a ditching or gear-up landing is at risk for serious injury due to the compression and/or collapse of the airplane structure. The NTSB notes that the A318, A319, A320, and A321 series airplanes have similar structures.

[ANALYSIS] The head and arms will slide down the seat back as it folds, but shouldn’t be seriously injured. The passenger seats on the accident airplane were 16-G compatible seats141 that had a nonbreakover seatback design, meaning that the breakover hinge feature was “locked out” and that the seatbacks were designed to be essentially rigid and to not easily or quickly collapse forward as passengers struck them from behind. All newly manufactured 16-G compatible seats have a nonbreakover seatback design, which minimizes head movement and body acceleration before striking the seatback from behind, resulting in less serious head injuries. The NTSB notes that the guidance in AC 121-24C did not take into consideration the effects of striking seats that do not have the breakover feature because research on this issue has not been conducted. The NTSB concludes that the FAA’s current recommended brace positions do not take into account newly designed seats that do not have a breakover feature and that, in this accident, the FAA-recommended brace position might have contributed to the shoulder fractures of two passengers. Therefore, the NTSB recommends that the FAA conduct research to determine the most beneficial passenger brace position in airplanes with nonbreakover seats installed. If the research deems it necessary, issue new guidance material on passenger brace positions. 2.10.2 Evacuation Flight attendant A reported no difficulties opening door 1L or getting it to lock against the fuselage; however, she reported that, although no water entered the door, the slide/raft did not inflate automatically. She stated that she pulled the manual inflation handle and that the slide/raft inflated normally. Although she did not report a delay in the inflation after pulling the manual 140 See R.F. Chandler, “Brace for Impact Positions,” Proceedings of the Fifth Annual International Cabin Safety Symposium, February 1988 Los Angeles, California (Los Angeles, California: University of Southern California, Federal Aviation Administration, and Southern California Safety Institute, 1988), pp. 279–290. 141 A 16-G seat is tested in a manner that simulates the loads that could be expected in an impact-survivable accident. Two separate dynamic tests are conducted to simulate two different accident scenarios: one in which the forces are predominantly in the vertical downward direction and one in which the forces are predominantly in the longitudinal forward direction. The highest load factor is in the forward direction at 16 G, which is why these seats are commonly referred to as 16-G seats. Amendment 121-315, effective October 27, 2005, required that transport-category airplanes in Part 121 operations, certificated after January 1, 1958, and manufactured on or after October 27, 2009, must comply with the 16-G dynamic standard. NTSB Aircraft Accident Report 106 inflation handle, postaccident video analysis indicated that about 20 seconds elapsed after she opened door 1L until the slide/raft began to inflate. Further, both of the first two passengers to reach door 1L reported that the flight attendant seemed to be struggling with the slide/raft.

[ANALYSIS] This conclusion is based on the fact that the flightcrew had several opportunities to prevent the accident. 2.9 Survivability According to the lead flight. attendant, seated in 2C, who survived with serious injuries, there was no warning to the cabin from the cockpit crew. regarding the low fuel status, loss of engines, or the impending emergency landing. Therefore, passengers were not briefed on _ brace positions, other than during the pretakeoff briefing, and on evacuation procedures. However, after the failure of all four engines and generators, the ability of the cockpit to communicate with the cabin on the PA system would have terminated. If the cabin crewmembers and passengers had assumed the brace position before the impact, the severity of some of the injuries might have been reduced. Seventy-two of the 74 passengers who survived sustained serious injuries. These injuries consisted of multiple lower leg fractures and dislocations, head injuries, hip fractures, spinal fractures, and multiple lacerations and contusions. ‘The legs of passengers probably impacted the lower seat back frames of seat units in front of them. Simultaneously, passenger seats most likely collapsed and twisted downward and to the left, resulting in hip and spinal fractures. As the impact sequence progressed, separation of the seat units from their floor attachments probably pushed passengers forward into other passengers, seat units, and other wreckage debris, causing head injuries and lacerations. Two of the 10 surviving infants sustained minor injuries, consisting of multiple contusions and abrasions and eight sustained serious injuries, consisting of limb fractures and head injuries. The 10 infants were either held by adult passengers or were belted into the same seat with the passengers. Surviving passengers who held infants reported that during the impact the infants were ejected from their grasp and that they were generally unable to locate them in the darkness after the impact. The Safety Board believes that the problems experienced in this and other accidents illustrate the impossibility of parents holding onto infants during a crash. If the infants had occupied FAA-approved restraint systems, injuries would most likely not have been as severe. The Safety Board addressed the subject of infant restraints in safety recommendations issued on May 30, 1990. They were: 72 Revise 14 CFR 91, 121, and 135 to require that all occupants be restrained during takeoff, landing, and turbulent conditions, and that all infants and smal] children below the weight of 40 pounds and under the height of 40 inches be restrained in an approved child restraint system appropriate to their height and weight. (Class I, Priority Action) (A-90-78) Conduct research to determine the adequacy of aircraft seatbelts to restrain children too large to use child safety seats and to develop some suitable means of providing adequate restraint for such children. (Class II, Priority Action) (A-90-79) The FAA Administrator responded to the recommendations on August 6, 1990.

[ANALYSIS] Regarding A-90-78, the FAA issued a Notice of Proposed Rulemaking (NPRM) on February 22, 1990, for child restraint system provisions. The Safety Board found the provisions in the NPRM unacceptable because they did not require the use of an approved restraint system. Rather, the provisions would merely prohibit) airlines from denying their use. Consequently, the Safety Board has classified A-90-78 as "Open--Unacceptabie Action." The FAA responded to recommendation A-90-79 that it has no research planned to examine adequate restraint systems for children who are too large to use child safety seats. Consequently, the Safety Board has classified this recommendation as "Open--Unacceptable Action." It could not be determined where all the passengers were seated at the time of impact, because the airline only assigned seats to a small percentage of passengers. Those passengers who had assigned seats stated that many of them had moved freely about the cabin to sit with family and friends. Therefore, passenger seat locations in relation to an individual injury diagram could not be developed with certainty in all cases. The captain, first officer, and flight engineer died from blunt force head and upper torso trauma. The captain and firs* officer seats had no shoulder harnesses installed. On March 6, 1980, the FAA required all cockpit seats to be equipped with combined seatbelts and shoulder harnesses; however, ICAO standards do not address these restraint systems. The right side of the cockpit struck a 42-inch-diameter oak tree that penetrated the space occupied by the first officer and the flight engineer, causing nonsurvivable injuries. Five of the six flight attendants were fat.ily injured as a result of blunt force injuries to the head, chest, abdomer, and limbs. Three of the five flight attendants’ locations could be established based on the statement of the surviving flight attendant. One was seated in the L-] jumpseat, the second was in passenger seat 2A, and the thirc was in passenger seat 3C. 73 Sixty-four adult passengers and one 4-month old infant died as a result of blunt force injuries. It is possible that some of the injuries could have been reduced, permitting some of these passengers to survive, if they had been instructed to assume the brace position before impact. Although there was no fire, it should bz noted that the cushions on the passenger seats were refurbished in March 1987, 9 months before the FAA regulation became effective requiring seat cushions to have fire-blocking material. ICAO Annex 8 recommends the fire-blocking material. Cabin floor proximity emergency escape path lights were not’ installed.

[ANALYSIS] The FAA responded that that most air carriers did not conduct joint emergency evacuation exercises and that the agency would not require such exercises. As a result, the NTSB classified Safety Recommendation A-00-85 “Closed—Unacceptable Action.” 93 Before this point, the lead flight attendant and another flight attendant had met in the middle of the airplane, and then the lead flight attendant went forward, and the other flight attendant went aft. Thus, it is not known if all flight attendants were off the airplane at that time. 94 The NTSB does not know how long it took for the captain to learn the occupant count after he called dispatch. 95 No company guidance indicated when a captain should not perform a walk-through of the cabin. However, the American Airlines manager of flight training and standards stated that a walk-through of the cabin should only be done if it would not compromise the safety of a captain. NTSB Aircraft Accident Report 65 After evacuating the airplane, the flight crew was responsible for helping assemble passengers away from the airplane, and the captain had the additional responsibility to ensure that at least one crewmember remained with the passengers. According to postaccident interviews, this was not done. Although the captain was unable to verify that everyone was off the airplane due to smoke in the cabin, the captain’s priority, after leaving the airplane, should have been to ascertain this information. However, the American Airlines B767 Operations Manual, QRH did not specify these postevacuation duties nor was there training that emphasized these duties. If these duties had been specified and trained, the captain might have been more likely to obtain an accurate passenger count from the flight attendants. Coordinating with flight attendants about the passenger count would have ensured, before ARFF personnel conducted its walk-though of the cabin, that all passengers were safely off the airplane or would have allowed ARFF to be promptly notified if the passenger count did not match the information provided by dispatch. The American Airlines B767 Operations Manual, QRH stated that a concurrent flight attendant evacuation duty, after leaving the airplane, was to make a count and report it to the captain. The NTSB found no evidence indicating that the flight attendants accomplished this duty. Even though this situation did not result in any adverse outcomes, the NTSB concludes that the flight crewmembers and flight attendants did not coordinate in an optimal manner once the passengers were evacuated. 2.3.4 Carry-On Baggage Issue Video taken during the evacuation and postaccident interviews with flight attendants indicated that some passengers evacuated from all three usable exits with carry-on baggage. In one case, a flight attendant tried to take a bag away from a passenger who did not follow the instruction to evacuate without baggage, but the flight attendant realized that the struggle over the bag was prolonging the evacuation and allowed the passenger to take the bag. In another case, a passenger came to the left overwing exit with a bag and evacuated with it despite being instructed to leave the bag behind. Passengers evacuating airplanes with carry-on baggage has been a recurring concern.

[ANALYSIS] In another case, a passenger came to the left overwing exit with a bag and evacuated with it despite being instructed to leave the bag behind. Passengers evacuating airplanes with carry-on baggage has been a recurring concern. In our June 2000 safety study on emergency evacuations of commercial airplanes, the NTSB stated that flight attendants had reported that their attempts to maintain a constant flow of passengers out an emergency exit “were often thwarted by passengers’ insistence on retrieving their carry-on luggage before evacuating.” Most recently, video from the British Airways event in Las Vegas and the Dynamic International Airways event in Fort Lauderdale showed passengers who evacuated with carry-on baggage despite the standard instruction to leave their baggage and similar items behind in the event of an emergency. Flight attendants are trained to instruct passengers not to evacuate with carry-on baggage because doing so could potentially slow the egress of passengers during an evacuation and block an exit during an emergency. Although the NTSB’s June 2000 safety study found that passengers exiting with carry-on baggage was “the most frequently cited obstruction to evacuation,” the NTSB has not identified any accident evacuations in which delays related to carry-on baggage NTSB Aircraft Accident Report 66 caused injuries. Also, the NTSB is not aware of any study performed to (1) measure the potential delays associated with passengers retrieving and carrying baggage during an emergency evacuation and (2) determine the appropriate countermeasures to mitigate any related potential safety risks. As a result of findings from our safety study on emergency evacuations of commercial airplanes, the NTSB issued Safety Recommendation A-00-88, which asked the FAA to “develop advisory material to address ways to minimize the problems associated with carry-on luggage during evacuations.” In January 2002, the NTSB classified the recommendation “Closed— Acceptable Action” because the FAA had revised Order 8400.10 to direct POIs to encourage their assigned certificate holders to provide, in crewmember manuals and training programs, clear direction and guidance to minimize the problems associated with passengers evacuating with carry-on baggage during emergencies. (This information was subsequently incorporated into FAA Order 8900.1.) In addition, AC 121-24C, “Passenger Safety Information, Briefing and Briefing Cards,” provides guidance to Part 121 carriers regarding the items that should be included in oral passenger briefings and on passenger briefing cards. Specifically, Appendix 1 states that oral briefings must be supplemented with briefing cards and that the content of the briefing cards should, among other things, “inform passengers that in an emergency situation, they should not bring carry-on baggage to the exit.” Further, AC 121-29B, “Carry-On Baggage,” states, in paragraph (k), that air carriers should provide training to all crewmembers regarding the carrier’s approved carry-on baggage program, including how to handle carry-on baggage during an emergency.

[ANALYSIS] The flight crewmembers then exited the cockpit and entered the forward cabin, which was filled with smoke, and were immediately met by the lead flight attendant, who told them that all passengers were off the airplane and that they needed to evacuate.93 The flight crewmembers and lead flight attendant then exited the airplane through the 1L door. ARFF personnel requested the occupant count from the captain, who called dispatch to get the count. ARFF personnel then walked through the cabin and confirmed that all passengers and crew were off the airplane. After the captain received the occupant count from dispatch, ARFF compared that number with the count of occupants who were off the airplane; everyone was accounted for.94 The American Airlines B767 Operations Manual, QRH instructed the captain to conduct a walk-through of the entire cabin to ensure that all passengers evacuated and then to exit the airplane through a usable aft entry door. However, the captain did not check the cabin to verify the lead flight attendant’s report that all passengers had exited the airplane, which was understandable because the captain reported that, upon exiting the cockpit, he could see “thick black smoke” and could not see farther than “about 2 feet” in front of him.95 92 Safety Recommendation A-92-74, issued as part of a special investigation report on flight attendant training and performance during emergency situations, asked the FAA to ensure that “all reasonable attempts” were made to conduct joint flight crew/flight attendant evacuation drills during recurrent training. In response, the FAA directed POIs to ensure that their assigned certificate holders were aware of the performance benefits that result when flight crews and flight attendants conduct emergency evacuation drills together. Because the FAA did not require air carriers to conduct joint exercises with flight attendants and flight crews, the NTSB classified Safety Recommendation A-92-74 “Closed—Unacceptable Action.” Also, as part of our June 2000 safety study on emergency evacuations of commercial airplanes, the NTSB issued Safety Recommendation A-00-85, which again asked the FAA to require air carriers to conduct periodic joint evacuation exercises involving flight and cabin crews. The FAA responded that that most air carriers did not conduct joint emergency evacuation exercises and that the agency would not require such exercises. As a result, the NTSB classified Safety Recommendation A-00-85 “Closed—Unacceptable Action.” 93 Before this point, the lead flight attendant and another flight attendant had met in the middle of the airplane, and then the lead flight attendant went forward, and the other flight attendant went aft.

[ANALYSIS] The AAIB’s report stated that the flight attendant who initiated the evacuation did not activate the evacuation signal (similar to the flight attendants aboard flight 383). However, another flight attendant went to the flight deck to report that an evacuation was underway, and the captain saw (from a reflection in the terminal building) that an aft emergency slide had been deployed. The captain then made an announcement to stop the evacuation because he thought that he had isolated the source of the smoke and wanted to prevent unnecessary injuries. However, the captain did not discuss the situation in the cabin with the flight attendants before making his announcement, which indicated “a breakdown in communication and co-operation between flight crew and cabin crew members.” The AAIB’s report also indicated that the captain’s announcement caused confusion. One of the flight attendants thought that the captain was not aware of the smoke in the cabin, so she shouted to the passengers to keep moving. Another flight attendant, who saw the captain standing in the flight deck, told the captain that the evacuation should continue because of “thick smoke” in the cabin, and the captain made a subsequent announcement indicating that the evacuation should continue via a jetbridge (which was in place before the evacuation). The AAIB’s report concluded that “prompt and effective communication between the cabin and the flight deck might have avoided an evacuation” and that one reason for the initiation of the evacuation was that the flight attendants “did not receive specific instructions from the flight crew.” NTSB Aircraft Accident Report 64 The AAIB noted that American Airlines had taken postincident actions in response to the communication and coordination shortcomings found during the investigation but that action was also needed by the regulator because other operators might be susceptible to similar shortcomings. As a result, the AAIB issued Safety Recommendation 2017-029, which asked the FAA to “require that flight and cabin crews participate in joint training to enhance their co-ordination when dealing with emergencies.” The NTSB has made similar recommendations to the FAA, but joint evacuation exercises for flight and cabin crews are still not required.92 The NTSB recognizes the benefits of joint flight and cabin crew evacuation training and notes that the actions requested in Safety Recommendation A-16-26 would also be an effective way to resolve evacuation-related communication and coordination issues. 2.3.3 Flight and Cabin Crew Evacuation Duties According to postaccident interviews with the flight crewmembers, they completed the evacuation checklist, including the step in which the captain announced over the PA system that an evacuation was underway.

[ANALYSIS] The first officer was wearing several layers of clothing and was thus afforded more protection from radiant heat and direct flames. None of the surviving passengers was incapacitated during the in-flight fire. The rapid evacuation of the passengers was possible because the cabin remained intact and no debris hampered access to the forward right emergency exit or to the opening in the rear of the cabin. Only one of the four emergency exits was used. Two additional exits could have been used if the passengers had read the safety card and been familiar with the emergency exits. It is noteworthy that of the nine surviving passengers, only the 9-year-old girl in seat 6A, which was directly above the cabin fire extinguisher, had read the safety card on this flight. Most of the passengers were not incapacitated by injuries sustained at impact; only three—the passengers in seats 2C, 6A, and 6C--had to be helped from the aircraft because of their injuries. The blind woman in seat 5A also was assisted from her seat to the exit. There was no reported panic or disruptive behavior among the passengers during the evacuation. The evacuation was rapid because the two exits could be reached quickly in the relatively small cabin even though visibility was reduced. The autopsy of the 59-year-old woman who did not escape revealed that she had severe chronic obstructive lung disease and severe cardiovascular disease. No bluntforce traumatic injuries were found. The elevated, though not necessarily lethal, level of carbon monoxide, 25 to 32.6 percent, and the 1.75 yig/ml level of hydrogen cyanide indicate respiratory activity in the presence of the byproducts of fire. The seat occupied by the victim exhibited slight deformation to the seatpan frame and the seatback frame. The man who occupied the seat directly behind her sustained a compression spinal fracture, and the seat he occupied exhibited damage which was consistent with his injuries. The man who occupied the seat directly in front of her received minor injuries, and his seat was one of three undamaged single seats. Thus, the lack of blunt-force injuries to the spinal column, head, and internal organs and the lack of major impact-related damage to her seat indicates that the woman who died was not subjected to erash forces of sufficient magnitude to damage her seat and inflict severe injury. Because the woman's body was found lying across the eabin from where she had been seated and because no seatbelt release buckle was found with the body, the Safety Board concludes that after the aircraft landed and after other passengers had a mee escaped, the woman attempted to escape but succumbed to the effects of hot gases and carbon monoxide before she was able to do so. 3. CONCLUSIONS

[ANALYSIS] The passenger seats were equipped for underseat stowage of carryon baggage in accordance with 14 CFR 135.87, but one passenger said he had his attache case on his lap during takeoff. Although 14 CFR 135.87(¢)(6) requires carryon baggage to be stowed before takeoff and before landing, there is no requirement that flighterews inform passengers of the requirement or that safety cards contain this information. Neither the pretakeoff briefing nor the safety briefing cards on the accident aircraft mentioned requirements for carryon baggage. This type of safety information becomes most important in commuter operations because operators usually are not required to have onboard a cabin attendant who would insure that the passengers have their seatbelts fastened, that there is no smoking, and that carryon baggage is stowed beneath seats. Consequently, the Safety Board believes that the pretakeoff briefing and safety briefing cards should be amended to inform commuter passengers of the carryon baggage stowage requirement. Title 14 CFR 135.117 requires that the pretakeoff briefing of passengers inelude information concerning the use of seatbelts. The Pilgrim Airlines flighterew pretakeoff briefing did not address seatbelt use. Likewise, the seatback safety briefing ecards did not address seatbelt usage or operation of the seatbelts as supplemental information. The Safety Board believes that the oral briefing should comply with the regulatory requirement and that the seatback safety briefing cards should provide supplemental information on the operation and the use of the seatbelts. The safety briefing card also did not show the location of the overhead escape hatch located in the rear cabin. The Safety Board believes that more comprehensive surveillance by FAA inspectors would have discovered the obvious discrepancies in the safety card and the lack of instructions for the use and operation of seatbelts in the pretakeoff briefing. The FAA's surveillance of air taxi and commercial operators which operate under 14 CFR Part 135 should certainly place heavy emphasis on occupant safety and safety equipment. The rapid growth of the commuter industry, as discussed in the Safety Board's special study on commuter airline safety, 5/ clearly requires that FAA inspectors be aware of the need for operators to conform to all applicable requirements in the Federal Aviation Regulations under which they operate. 2.4 Survival ets Fire injuries were sustained by the captain, first officer, and two passengers. The crew sustained their injuries in flight and before escaping from the aircraft after it landed. The captain's injuries were more severe and extensive because the fire appeared to have been more extensive in the left cockpit and left forward cabin. The first officer was wearing several layers of clothing and was thus afforded more protection from radiant heat and direct flames. None of the surviving passengers was incapacitated during the in-flight fire. The rapid evacuation of the passengers was possible because the cabin remained intact and no debris hampered access to the forward right emergency exit or to the opening in the rear of the cabin.

[CONCLUSIONS > FINDINGS] Passengers 41B and 41E were unrestrained for landing and ejected through the ruptured tail of the airplane at different times during the impact sequence. It is likely that these passengers would have remained in the cabin and survived if they had been wearing their seatbelts. NTSB Aircraft Accident Report 128 19. Passenger 42A was likely restrained for landing, and the severity of her injuries was likely due to being struck by door 4L when it separated during the airplane’s final impact. 20. The dynamics of the impact sequence in this accident were such that occupants were thrown forward and experienced a significant lateral force to the left, which resulted in serious passenger injuries that included numerous left-sided rib fractures and one left-sided head injury. 21. The reasons for the high number of serious injuries to the high thoracic spine in this accident are poorly understood. 22. The release and inflation of the 1R and 2R slide/rafts inside the airplane cabin was a result of the catastrophic nature of the crash, which produced loads far exceeding design certification limits. 23. Clearer guidance is needed to resolve the concern among airport fire departments and individual firefighters that the potential risk of injuring airplane occupants while piercing aircraft structure with a skin-penetrating nozzle outweighs the potential benefit of an early and aggressive interior attack using this tool. 24. Medical buses were not effectively integrated into San Francisco International Airport’s monthly preparation drills, which played a part in their lack of use in the initial response to the accident and delayed the arrival of backboards to treat seriously injured passengers. 25. Guidance on task prioritization for responding aircraft rescue and firefighting personnel that addresses the presence of seriously injured or deceased persons in the immediate vicinity of an accident airplane is needed to minimize the risk of these persons being struck or rolled over by vehicles during emergency response operations. 26. The overall triage process in this mass casualty incident was effective with the exception of the failure of responders to verify their visual assessments of the condition of passenger 41E. 27. The San Francisco Fire Department’s aircraft rescue and firefighting staffing level was instrumental in the department’s ability to conduct a successful interior fire attack and successfully rescue five passengers who were unable to self-evacuate amid rapidly deteriorating cabin conditions. 28. Although no additional injuries or loss of life were attributed to the fire attack supervisor’s lack of aircraft rescue and firefighting (ARFF) knowledge and training, the decisions and assumptions he made demonstrate the potential strategic and tactical challenges associated with having nonARFF-trained personnel in positions of command at an airplane accident. NTSB Aircraft Accident Report 129 29. Although some of the communications difficulties encountered during the emergency response, including the lack of radio interoperability, have been remedied, others, such as the breakdown in communications between the airport and city dispatch centers, should be addressed. 30.

[ANALYSIS] Passengers in this section sustained nonsurvivable blunt trauma injuries. Rows 9 through 14 and the left wing were located in the front yard of a private residence and had extensive fire damage. In this section, some passengers sustained fatal bluat force injuries, some others died of the effects of the fire, while others died from a combination of blunt force injuries and the fire. Two passengers, although injured, were able to escape before the fire intensified. Passengers in rows 15 and 16 sustained blunt force and/or thermal injuries. The aft section of the airplane, which included seat rows 17 through 21, was found imbedded in the carport of the residence. Occupants of this section sustained blunt force injuries and burns. Some of the blunt force injuries occurred to occupants of seat rows 17 and 18 when those rows rolled under the tail section during the impact sequence. The Safety Board believes that the more serious bum injuries sustained by these survivors occurred as they escaped the left side of the wreckage. Based on their injuries, the Safety Board believes that the passengers who were assigned seats 20E and 20F were not in those seats at the time of the accident. There were 57 occupants on flight 1016: 52 passengers and 5S crewmembers. The passenger manifest listed 50 names and did not include the names of two in-lap infants. Federal Aviation Regulations specifically address the issue of passenger manifests. 14 CFR 121.693(e) requires that a load manifest contain the names of passengers aboard the airplane. The FAA also issued Action Notice 8340.29 and Air Carrier Operations Bulletin No. 8-91-2 to reaffirm that every occupant, who is not a crewmember with assigned duties, must be listed on the passenger manifest. The USAir procedures for accounting for in-lap infants requires the gate agent to place an "Infant Boarding Pass-Non Assigned Seat" sticker and the remark "Plus Infant’ in the name field of the accompanying adult's flight coupon. Neither of the flight coupons for the adults associated with the two in-lap infants included an "infant boarding pass" sticker. Although one coupon included a handwritten notation "+ infant,” the second coupcn did not; thus, the infant svas not included on the passenger manifest. The Safety Board identified inaccuracies :vith passenger manifests in several previous accident investigations 2nd issucd Safety Recommendations A-79-65 and A-90-105 that asked the FAA to require standardized reporting by air carriers of passengers on manifests. Both of these recommendations have previously been classified as “Closed--Acceptable Action.” Whiie the FAA has established rules that there be an accurate listing of occupants (14 CFR 121.693(e)) regarding manifests, and USAir has procedures for accounting fcr in-lap infants, these procedures are not consistently followed. The Safety Board believes that USAir should review its procedures to ensure that manifests have an accurate count of all occupants on each airplane.

[ANALYSIS] Most of the passengers who received injuries were in thelr seats with the seatbelts elther unfastened or fastened loosely. tnsetructions contained in the National Airlines Stewardess Manutl ragarding "Regulations" state: Whenever the seatbelt sign is turned on, the appropriate (Turbulence or Descending) announcement must be made. Seatbelts must be checked. Accordingly, the Board concludes that preoccupation with en attempt to provide acceptable meal service and an extensive timo pertod (30 minutes ) of seatbelt sign Lllumination prevented the coach section stewardesses from maintal.ning effective seatbelt discipline in the two aft coach sections. Because of the reported deficiencies regarding Pirst-ald supplies onboard WA 4L}} in particular, regarding adhesive tape, large gauze compresses, larga triungular bandages, and material for splints, the Beard is concermed that emergency medieal supplies required on large commerical atreraft are inadequate in poth content and quantity. As the number of passengers Lnereases, 60 should the required number of Cirste ald supplies to provide properly for in-flight situations which may arise. uikewlse, the first-aid Kit contents should be broadened to Include those items requirea to care for injuries of a nore perilous neture. PROBAGLE CAUSE The Notional Transportation Safety Poard deternines that the probable cause of this seeldent was an encounter with sharp-gust cone vective turbulence during flight in instrumert meteorological conditions while mumerous occupants of the alreraft were unsccured by seatbelts even though the seatbelt sign was lighted. The Board also determines that 9, number of passengers were injured because priority was given by the stoewardesses to vegular passenger service duties rather than to the entorcenent of seatbelt usage. I pee a et ene AS IEEE 2 ANE Bed FEET OE oe peng spcciarwetnige s Winns weal RSA ta RRR Se mse emer - 8 ~ RECOMMENDATIONS eek eh erenanens manemennh . wiitine Tenth Subsequent to the investi gueton of « previous B-747 accident involving turbulence, the Netional Transportation Safety Board recommended to the Federal Aviation Administration improvements or corrective action concerning seatbelt discipline and alr carrier policy on deviation of flight. : As a result of this investigation,on April 19, 19/2, the Beard recommended thet the Federal Aviation Adwninistratio..: (1) reevaluate the installation of Sundetreand recorder on B-7'7? aircraft, and (2) issue eau Airworthiness Directive requiring modificetion of the Sundstrand recorder vertical. accelerometer system to conform to the Federal Aviation Regulations. Oo On June 14, 1972, tr: Board further recommended that the Federal.

[ANALYSIS] ANALYSIS Except for a 3-hour takeoff delay becausa of maintenance difficulties, the flight of NA hi was routine witil lts encounter with turbulence southeast of Lake Charles, Louisiana. Weather radar photographs taken at Lake Charles, Loulslana, within L minute of the estimated accident time, showed an extensive area of weather activity to the southeast of Lake Charles at a distance of 50 ta 100 miles. Accordingly, the Safety Board concludes that, even though the alrcratt was clear of the heaviest precipitation areas depicted on the aircraft's radar at the time of the encounter, the aircraft was within the influence of the strong vertical wind currents commonly found in and around thunder« storm activity. Subsequent to the accident, the flight data recorder unit was sub jected to tests by The Boeing Company to verify ite ability to respond to charac~- teristic inputs. The vertical- acceleration channal was found to have exces~- sive damping. Th> Board concludes that the recorder failed to respond to the pertinent inputs, and further, that the unit 2s installed did not comply with Sections 121.343 and 37.150 of the Federal. Aviation Regulations. The absence of a valid vertical-acceleration trace on the Clight data recorder preciuded a determination of the magnitude and time history of the forces acting on the aft portion of the fuselage In the urea where the passengers and stewardesses were injured. A yeview of injuries and statements of unrestrained passengers and stewardesses disclosed that the aft section was subjected to an acceleration acting downward and to the left. According to crewmember and passenger statements, the seatbelt 8 J.gn had been turned on approximately 30 minutes prior to the turbulence encounter. However, as the flight hed remained smooth prior to the jolt, the captain had no reason to instruct the stewardesses to suspend meal service and be seated. Consequently, four stewardesses Located In ‘the aft coach sections were preparing meal. service when the jolt was encountered. This aceldent vividly Wustrates the risk cabin attendants repularly maz contend with when they are engaged in normal cabin duties and turbulence is encountered. 26 Sans ORE bad srs ya aes iF vp etary, carne ghee D We LSet yi cee NE FEILER LE NEA YET SF we ae eS e maretariey cna ees enc ETE nes ened -T- ALL injuries were sustained by passengers and eremmerbers in the two af}} coach sections of the aircraft. Most of the passengers who received injuries were in thelr seats with the seatbelts elther unfastened or fastened loosely. tnsetructions contained in the National Airlines Stewardess Manutl ragarding "Regulations" state: Whenever the seatbelt sign is turned on, the appropriate (Turbulence or Descending) announcement must be made. Seatbelts must be checked.

[ANALYSIS] Further, regulations prohibit taxiing, taking off, or landing when any beverage provided by the operator is located at any passenger seat.97 The cabin aide was not required to receive any safety-related training because she was not a required crewmember (flight attendant) for the accident flight. Nonetheless, PJM did provide its cabin aides with some training. The accident cabin aide stated that she had received verbal instruction regarding emergency main cabin door operation and had operated the main cabin door handle and electric toggle switch in a simulated emergency scenario during her training. Her description of her efforts to operate the door after the accident was consistent with the training she reported that she had received; however, it revealed that her training had not provided her with an adequate understanding of the door mechanism/operation. The cabin aide told investigators that she was not familiar with the arm/disarm handle and that she tried to use the electric switch at the top of the bulkhead to operate the door; however, this switch is not needed (nor should it be used) during emergency operations. The Safety Board is concerned that Part 135 operators and/or certificate holders may delegate important safety functions to cabin aides/customer service representatives who are not properly trained and qualified to perform those functions. For example, although PJM policy required one of the pilots to provide passengers with a safety briefing before each flight, the captain believed that the cabin aide was responsible for the preflight safety briefing.98 Further, the Board is concerned that passengers might mistakenly believe that a cabin aide/customer service representative on board an on-demand charter flight had received safety training equivalent to that received by qualified flight attendants when in fact that aide/representative might have received minimal or no safety training. Providing those individuals with basic safety training could provide valuable safety results in an emergency, especially in the event of flight crew injury. 96 Title 14 CFR 135.128 states, “each person on board an aircraft operated under this part shall occupy an approved seat or berth with a separate safety belt properly secured about him or her during movement on surface, takeoff, and landing.” 97 For additional information, see 14 CFR 135.295, 135.117, 135.122, and 135.128. 98 The captain had not yet received PJM policy training. Analysis 61 Aircraft Accident Report On the basis of the cabin aide’s performance during this accident sequence, including the lack of a seatbelt compliance check, her failure to collect beverage service items before takeoff, and her inability to open the main cabin door and conduct a professional evacuation, the Safety Board concludes that the cabin aide’s training did not adequately prepare her to perform the duties with which she was tasked, including opening the main cabin door during emergencies. Therefore, the Safety Board believes that the FAA should require that any cabin personnel on board 14 CFR Part 135 flights who could be perceived by passengers as equivalent to a qualified flight attendant receive basic FAA-approved safety training in at least the following areas: preflight briefing and safety checks; emergency exit operation; and emergency equipment usage.

[ANALYSIS] For example, after the passengers boarded the airplane, the cabin aide offered the passengers a beverage and four accepted. The drinks were served in glasses or ceramic/china cups, several of which were recovered on or near passenger seats after the accident, which is not consistent with a secured cabin. One passenger told investigators that he had to pick up his coffee cup during the takeoff roll to prevent spillage. He believed that the lacerations on his right hand were caused by the coffee cup breaking in his hand during the accident sequence. Additionally, postaccident interviews indicated that at least four of the eight passengers were unrestrained when the takeoff roll began. Two passengers located and fastened their seatbelts during the takeoff roll; however, the other two (both seated on the side-facing divan seat) were unable to locate their seatbelts and were therefore unrestrained during the RTO. The two unrestrained passengers were thrown to the cabin floor during the accident sequence. Postaccident examinations revealed that the seatbelts at the three divan seats would not have been visible to the passengers because they had been intentionally placed beneath the seatback cushions. Postaccident examination of another PJM CL-600 revealed that its divan seatbelts had also been intentionally placed beneath the seatback cushions. Positioning the seatbelts beneath the seatback cushions resulted in a tidier passenger cabin and was reportedly not uncommon among operators of corporate and charter airplanes. However, with the seatbelts stowed in this position, passengers would have had to either reach blindly between the seatback cushions or remove the cushions to locate the seatbelts. The Safety Board concludes that the cabin aide did not perform a seatbelt compliance check before the accident flight, which resulted in two passengers being Analysis 60 Aircraft Accident Report unrestrained during the accident sequence. Further, the Safety Board concludes that the intentional positioning of the seatbelts out of the passengers’ sight made them difficult to locate and use and resulted in reduced compliance with passenger seatbelt usage requirements. Therefore, the Safety Board believes that the FAA should require all 14 CFR Part 135 certificate holders to ensure that seatbelts at all seat positions are visible and accessible to passengers before each flight. According to Federal regulations, no person may serve as a “flight attendant” on a Part 135 flight unless that person is knowledgeable and competent in the areas of crewmember functions and responsibilities during ditching and evacuation, briefing of passengers, the location and operation of all normal and emergency exits, portable fire extinguishers, and other items of emergency equipment. Additionally, regulations require that passengers on airplanes certificated to carry 19 passengers or less must be briefed on, among other things, the use of safety belts96 and the location and operation of the main cabin door and emergency exits. Further, regulations prohibit taxiing, taking off, or landing when any beverage provided by the operator is located at any passenger seat.97 The cabin aide was not required to receive any safety-related training because she was not a required crewmember (flight attendant) for the accident flight.

[ANALYSIS] The ARFF response from TEB and surrounding communities was prompt, and exterior postaccident fires were extinguished efficiently. A TEB-based HRET/skin-penetrating-nozzle-equipped vehicle, had one been available, would have reached the accident site to extinguish the interior fire more promptly than the EWR-based HRET/skin-penetrating-nozzle-equipped vehicle that responded to this accident site. However, because all airplane occupants had self-evacuated before ARFF personnel 89 The ASOS does not provide direct information regarding frost, but indirect analysis of the data can determine whether frost was present or not. Analysis 48 Aircraft Accident Report arrived, earlier arrival of such a vehicle would not have facilitated passenger evacuation in this case. The TEB local controllers observed that the airplane was not pitching up at what they believed to be a “normal” liftoff point. Recognizing that the airplane would probably not be able to stop on the remaining runway, the controllers alerted the airport ARFF facility of the situation. The local controllers also promptly issued go-around instructions to an inbound airplane and coordinated that go-around and the cessation of further arrivals at TEB with the approach control facility. Further, the TEB ground controllers ensured that the ARFF responders were aware of the accident location and monitored their progress to the accident site. The Safety Board concludes that the air traffic controllers’ prompt and efficient reaction to this accident was exemplary and facilitated the prompt ARFF response. All eight of the passengers and the cabin aide received minor injuries (including contusions, abrasions, lacerations, sprains, and strains) during the impact and evacuation. The flight crew’s injuries (which included fractures, dislocations, and lacerations) were more serious but were mainly limited to the lower limbs. All of the passengers and crew evacuated through the main cabin door, and no attempt was made to use the overwing exit on the right side of the airplane. (Use of the overwing exit would have been ill-advised because of its proximity to burning vehicles.) Because the pilots were still trapped in their seats and the cabin aide was unable to open the main cabin door, passengers rotated the door handle and pushed and kicked the door open. A passenger reported that he fumbled to find the door handle in the postimpact darkness; however, the main difficulty opening the door was the result of the door being jammed. Postaccident investigation revealed no evidence of a preimpact door anomaly. The Safety Board concludes that doorframe and/or fuselage distortion caused by impact with the building accounted for the passengers’ difficulty opening the main cabin door. 2.2 The Accident Sequence The captain was the flying pilot during the takeoff roll/RTO. Both pilots indicated that the flight controls operated freely and normally during the preflight control checks and that no anomalies were noted before the takeoff roll was initiated about 1 minute later. Witness reports, statements from the pilots and passengers, airplane performance calculations, CVR sound spectrum analysis, and airport surveillance videotapes all indicated that the airplane accelerated for takeoff normally.

[CONCLUSIONS > FINDINGS] An earlier problem with fumes in the passenger cabin contributed to the cockpit crew's failure to evaluate correctly the cause of the in-flight problem. 16. No in-flight emergency was declared. 17. No further action could have been taken by the cockpit crew to land the airplane more quickly after the in-flight problem was detected. 18. Because no in-flight emergency was declared and the airplane was not evacuated immediately after landing, passengers and crewmembers were unnecessarily exposed to the threat of fire for 1 1/2 minutes longer than necessary, and the flight attendants were not able to prepare the passengers before landing for a quick exit. 19. Emergency passenger evacuation procedures on passenger safety cards were not consistent with oral instructions. 20. Air carrier ground personnel increased the risk of spreading the cargo compartment fire by opening cargo compartment doors. 21. Air carrier ground personnel unnecessarily risked personal injury by entering the passenger cabin of the airplane to remove passenger belongings before the fire was extinguished and the airplane determined to be safe. 3.2 Probable Cause The National Transportation Safety Board determines the probable cause of the in-flight fire to be a chemical reaction resulting from a hydrogen peroxide solution, in concentration prohibited for air transportation, which leaked and combined with the sodium orthosilicate-based mixture from an undeclared and improperly prepared container. The probable cause of the unauthorized transportation was the shipper's lack of knowledge about restrictions and requirements for hazardous rnaterials and inadequate procedures for detecting undeclared hazardous materials shipments. Contributing to the delay in detecting the in-flight fire and the captain's decision not to declare an in-flight emergency was the lack of heat or smoke detection equipment in the cargo compartment and insufficient flightcrew communication. Contributing to the threat to the airworthiness of the airplane was the lack of a fire extinguishment system for the cargo compartment and the inadequate design of the cargo compartment ceiling.

[ANALYSIS] That is, his current flight path, speed, and traffic sequence already was directed toward getting the airplane on the ground expeditiously, and he considered an expeditious landing the only immediate option available to alleviate this abnormal and ill-defined situation. The Safety Board believes that these circumstances may have operated in concert to precispose the captain to disbelieve the reports of smoke, and to establish a mind set that the cabin crew was instead experiencing the less serious fumes. The captain’s skepticism about the report of smoke was also reflected in the first officer's dialogue with the cabin crew His cornments appear to be mere of a challenge of the accuracy of the reports than an effort to get additional details. Even atter Fe determined the problem in the cabin to be serious and after he recognized the need for timely firefighting assistance on landing, the first officer failed to aggressively recommend that crastVfire/rescue equipment meet the airplane. On identifying smoke in the passenger cabin, flight attendant No. 4 recognized the potential seriousness of the problern and without hesitation, even under “sterile cockpit” conditions, immedrately informed the first officer about the condition. Subsequent actions by the cabin crew, including efforts to locate the source of the fire, maintaining open communications with the cockpit, using a deadheading crewmember to evaluate and communicate information about the problem, and moving passengers from the affected area, also demonstrated that they considered the problem to be serious in conclusion, the Safety Board believes that while it is untikely that the captain could have taken any action to land the plane more quickly, the cockpit crew failed to use the cabin crew effectively to obtain an accurate understanding of the developing problem. Had communications between the cockpit crew and the cabin crew been more effective, the Safety Board believes that the captain would have called for fire/rescue equipment to meet the airplane and ordered an emergency evacuation on the runway. The Safety Board betieves that American Airlines should use this exarnple in cockpit and cabin crew coarcination training to illustrate the need for cockpit crews to more effectively use cabin crews in describing suspected in-flight safety problems and to emphasize the need for cabin crews to be assertive when communicating information about safety problems to cockpit crews. The Safety Board previously zddressed the issue of cockpit and cabin crew coordination training as a result of its investigation of the in-flight fire aboard a OC-9 at Cincinnati, Ohio, on June 2, 1983.?2 As a result of its investigation, the Safety Board issued Safety Recommendation A-84-76 which called for the FAA to require its principal operations inspectors to review air carrier training and if necessary, require amendments concerning actions flight crews should take for immediately and aggressively determining the source and severity of any reported cabin fire. In responses to this recommendation, on November 2, 19&4, anc March 7,1986, the FAA advised that it believed that current rules and quidance did not warrant further action.

[ANALYSIS] Arecraft Accident Report--Air Canada Fight 797 McDonnell Douglas OC-9-32, C-FTLU, Greater Cincinnats International Airport, June 2, 1983 (NISB/AAR-8409). The Safety Board concluded that the actions of the flight attendants were performed in accordance with American Airlines training and procedures. The Safety Board noted that American Airlines emergency procedures require flight attendants to instruct passengers to remove shoes, while passenger safety information cards provide no similar instructions. The Safety Board believes that the communication of emergency evacuation procedures to passengers could be improved if American Airlines operational procedures, manuals, training, the flight attendants’ oral instructions, and passenger safety information cards provide consistent instructions to passengers regarding the removal of shoes. The Safety Board also urges the FAA to instruct principal operations inspectors to determire if passenger safety cards and flight attendant instructions to passengers for emergency evacuations are consistent with each air carrier's evacuation procedures. Although some air carriers instruct passengers to remove shoes during unplanned emergency evacuations to prevent damage to slidus, other air carriers do not. The Safety Board is aware that slide manufacturers have not recommended that shoes be removed. Certification demonstrations by air carriers and airplane manufacturers of evacuation systems have been routinely conducted with persons wearing tennis-type shoes and other low-heeled shoes Although there have been instances when passengers’ snoes, particularly women’s high-heeled shoes, have damaged stides or have caught on the slide fabric and injured persons; these instances are infrequent. On the other hand, there have been instances when passengers and crewmembers have removed shoes and successfully evacuated a crashed airplane only to sustain frostbite and injuries when they watked on wreckage and through fire. The Safety Board is also aware of recent actions by the FAA to require the sliding surface of evacuation slides to be more puncture resistant. It appears that in view of the FAA’s recent actions and the need for the crew and passengers to have foot protection following an evacuation, the FAA should research the safety aspects of removing shoes during an evacu tion. After the airplane was evacuated, actions taken oy American Airlines ground personnel, although well intended, could have resu!ted in the destruction of th.e airplane or the loss of lives. By opening the doors to cargo compartments suspected to contain fires without having the appropriate firefighting equipment available, ground personnel may compromise cargo compartment fire safety systems, supply oxygen to fires, and cause fires to spread or intersify. Ground personne! who are expected to respond to an aircraft when a fire is suspected shouid be trained on the appropriate actions to be taken. Further, airline personne! should be instructed not to board aircraft to collect the passengers’ carry-on luggage until the aircraft has been declared safe by fire personnel. 31 3. CONCLUSIONS

[FINDINGS] The flight crew’s continued descent to Philadelphia International Airport was not 4. inappropriate given that there was no evidence of abnormalities other than the odor and that no cockpit alerts had activated. The increased airflow that resulted from the Fumes Evacuation checklist actions 5. diluted the smoke and inhibited its detection by either the smoke detection system or flight crewmembers and provided the fire with additional oxygen. The aviation industry initiative on smoke, fire, and fumes provides specific guidance 6. on when and how flight crews should respond to evidence of a fire in the absence of a cockpit smoke and/or fire warning. The fire on board the accident airplane initiated as a smoldering fire. 7. The fire was detected by the airplane’s smoke and fire detection system after the fire 8. breached a cargo container, at which time, it proceeded to spread, and the growth of the fire after landing was fed by air entering through open doors and burnthrough holes. The exact origin and cause of the in-flight fire on board the airplane could not be 9. determined due to the destruction of potentially helpful evidence; however, the available evidence suggests that the fire most likely originated in container 12, 13, or 14. The current certification test standards and guidance for smoke or fire detection 10. systems on board many aircraft are not adequate because they do not account for the effects of cargo and cargo containers on airflow around the detection sensors and on the containment of smoke from a fire inside a container. Conclusions National Transportation Safety Board A I R C R A F T Accident Report 65 The threat from cargo fires could be mitigated by the installation of fire suppression 11. systems. Flight crews on cargo-only aircraft remain at risk from in-flight fires involving both 12. primary and secondary lithium batteries. The emergency response for this accident was timely. 13. Some aircraft rescue and firefighting personnel are not adequately trained on the 14. use of the high-reach extendable turret with skin-penetrating nozzle, reducing the effectiveness of the device in fighting interior aircraft fires. Philadelphia International Airport aircraft rescue and firefighting personnel were not 15. familiar with the accident airplane’s main cargo door, which adversely affected their ability to access the airplane’s interior to fight the fire. The availability of accurate and complete airplane diagrams would improve aircraft 16. rescue and firefighting personnel’s knowledge and familiarity with fleet configurations and would facilitate emergency response operations. A floor level emergency exit, including one equipped (when appropriate) with an 17. evacuation slide, would enable more efficient emergency egress for airplane occupants than cockpit window exits, and the associated, instructional placarding of such an exit would assist emergency responders with locating and operating the exit door and accessing the interior of the airplane. United Parcel Service Company (UPS) guidance on hazardous materials information 18. retrieval and dissemination was inadequate, which resulted in UPS personnel not providing emergency responders with detailed information about the hazardous materials on board the airplane in a timely manner. The requirements of 49 19.

[ANALYSIS] However, the flight engineer saw no smoke either of the two times that he visually checked the main cargo compartment. This evidence indicates that the fire initially did not generate a significant amount of smoke and was most likely initiated as a smoldering fire inside a cargo container. The construction of the cargo containers, which results in restricted airflow in or out of the container, likely inhibited the growth and detection of the fire in its initial stages. On the basis of this evidence, the Safety Board concludes that the fire on board the accident airplane initiated as a smoldering fire. Once the fire breached the cargo container in which it initiated, it would have begun to spread to adjacent containers. Detection of the fire by the main cargo compartment smoke detectors most likely occurred around the time of the first container burnthrough. The smoke detector in lower cargo compartment 33 alerted about 1 minute after the main cargo compartment alert, and some of the captain’s displays then began to falter, indicating the continued progression of the fire. The flight engineer first saw smoke when he exited the cockpit to close the main cargo air shutoff valve and black smoke emanated from the valve’s access panel. About 2 minutes later, almost immediately after touchdown, the flight engineer reported that smoke had begun entering the cockpit. The smoke continued to worsen after the airplane came to a stop, and the smoke in the cockpit became so thick that the two pilots could not see each other before evacuating the airplane. ARFF personnel who entered the airplane through the L1 door observed smoke but no fire in the main cargo compartment. Flames were first observed about 40 minutes after the airplane landed when ARFF personnel opened the right-forward overwing hatch Analysis National Transportation Safety Board A I R C R A F T Accident Report 50 and noticed flames above the containers just aft of the opening. The continued growth of the fire after the airplane landed was probably affected by the introduction of fresh air through the openings in the airplane, both through its open doors and eventually through holes created in the fuselage as the fire ultimately burned through the fuselage crown. The growth of the fire was also affected by the combustible nature of the packages and packing materials in the cargo containers, which provided a readily ignitable fuel source for the growing fire. The fuselage burned through aft of the wings (near containers 12, 13, and 14) about 2 hours after landing. On the basis of this evidence, the Safety Board concludes that the fire was detected by the airplane’s smoke and fire detection system after the fire breached a cargo container, at which time, it proceeded to spread and that the growth of the fire after landing was fed by air entering through open doors and burnthrough holes. Origin and Cause of Fire 2.4  Despite the length of time that the fire burned and the resulting destruction of potentially helpful evidence, the postfire condition of the cargo containers and contents and the surrounding airplane structure were examined for any evidence indicating the location from which the fire initiated.

[ANALYSIS] The lowest point of fuselage damage, which could be indicative of the point of origin of the fire, was located in the right aft portion of container 12 near container  13, where the damage extended down to the cargo compartment floor. Container 14 exhibited heavy damage to the contents and was relatively near the areas of the fuselage with the most damage. In addition, a large number of the items in this container were not identified either during the on-scene examination or through the shipper interview calls. As a result, container 14 could not be eliminated as a source of the fire. No evidence of explosion or high-temperature fire (for example, melted steel components) was found. Several factors eliminated the airplane’s wiring as a possible source of the fire, including the lack of any electrical burning odor, the lack of fused wires or damage associated with arcing wires, and the lack of system or display anomalies until almost 20 minutes after the flight crew first detected an odor. Other than secondary lithium batteries, which are discussed below, all of the recovered declared and undeclared hazardous materials were eliminated as ignition sources because they were found intact and relatively undamaged. No additional hazardous or questionable items (that is, undeclared hazardous, chemical, or flammable materials) were identified during the identification effort. Examinations of the cargo containers and their contents revealed that a number of electronic devices, such as laptop computers, contained some type of power source, specifically secondary lithium cells. All electronic devices and battery packs and cells that had any type of thermal damage were examined on scene; however, most of the items were too damaged for further documentation. Several laptop computers, loose battery cells, and battery packs with severe damage were retained and sent to the Safety Board’s laboratory for further examination. During the cargo examinations, no batteries were found that exhibited any damage identifying a source of ignition. Because the fire most likely originated in cargo container 12, 13, or 14, investigators attempted to determine the contents of the packages that were not accounted for on scene by contacting the shippers of the packages in these containers. This effort was unable to determine the contents of all of the packages in these containers; however, the effort did reveal that several electronic devices likely containing secondary lithium batteries were shipped in these containers. Unfortunately, the lack of information about the devices or the batteries prevented any determination of whether these batteries were associated with previously known recalls. The Safety Board concludes that the exact origin and cause of the in-flight fire on board the airplane could not be determined due to the destruction of potentially helpful evidence; however, the available evidence suggests that the fire most likely originated in container 12, 13, or 14.

[ANALYSIS] Origin and Cause of Fire 2.4  Despite the length of time that the fire burned and the resulting destruction of potentially helpful evidence, the postfire condition of the cargo containers and contents and the surrounding airplane structure were examined for any evidence indicating the location from which the fire initiated. As a result of these examinations, the lower cargo compartments were eliminated as the origin of the fire because they sustained no heat‑related damage, and the outer surfaces of their ceiling liners only exhibited sooting and smoke damage. The fire/smoke detector activations in lower cargo compartments 33 and 34 resulted from smoke infiltrating the lower compartments from the main cargo compartment above this area. Cargo containers 1 to 11 were eliminated as the fire initiation location because they sustained minimal to moderate thermal damage and the surrounding fuselage sustained little to no structural damage. Further, the ceiling liners were still in place for containers 1 to 7. Although the ceiling liners in containers 8 to 11 sustained thermal damage, it most likely resulted after the overwing hatches were opened. Containers 15 and 18 were eliminated as the fire initiation location because the cargo in these containers was accounted for during the on-scene examination and no evidence of ignition sources was found. The contents in containers 16 and 17 were heavily damaged, and fuselage burnthoughs were located near both containers. However, containers 16 and 17 were eliminated as the fire initiation location, in part, because the aftward movement of air in flight and on the ground would have made it unlikely that the fire started in either of these containers. Further, the amount of damage sustained near cargo container 17 might have resulted when ARFF personnel opened the L4 and R4 doors soon after they arrived on scene. ARFF personnel indicated that no fire was observed inside the rear of the airplane at this time. Pictures taken soon after the response began do not show any visible fire in that area, but pictures taken 2 hours into the emergency response show this area heavily involved in fire. Analysis National Transportation Safety Board A I R C R A F T Accident Report 51 Conversely, ARFF personnel did report first observing flames inside the airplane in an area near containers 12 and 13, and this area is also the area above where the first smoke detector activated in the lower cargo compartments. Further, the amount of damage to containers 12, 13, and 14, their contents, and the areas surrounding these containers, including two complete and a partial fuselage burnthrough, indicates that the fire most likely originated in one of these cargo containers. The lowest point of fuselage damage, which could be indicative of the point of origin of the fire, was located in the right aft portion of container 12 near container  13, where the damage extended down to the cargo compartment floor. Container 14 exhibited heavy damage to the contents and was relatively near the areas of the fuselage with the most damage.

[ANALYSIS] Analysis National Transportation Safety Board A I R C R A F T Accident Report 48 Use and Adequacy of Smoke, Fire, or Fumes Checklists 2.2.3  During the review of the flight crew’s actions and decision-making, investigators found that company guidance and checklists regarding smoke, fire, or fumes in the absence of a cockpit warning were not adequate because neither UPS nor Boeing provided specific flight crew procedures for responding to such a situation. Instead, the UPS DC-8 AOM provided four checklists that could have been applied to such a situation, three of which were predicated on visible evidence of smoke or fire or an alert activation in the cockpit. As noted, during this period, no smoke or fire warning lights illuminated in the cockpit, no visible evidence of smoke or fire existed, the CVR recorded no comments by the flight crew about burning eyes or headaches, and no evidence of abnormalities other than odor existed. Although one of the checklists, the Fumes Evacuation checklist, did apply specifically to fumes, the AOM did not provide guidance on when to use the checklist. Increasing both air conditioning packs to maximum flow (as part of the Fumes Evacuation checklist) would have increased the airflow through the cabin. This would have evacuated the smoke more quickly, diluting the air and inhibiting the flight crew’s ability to identify the source of the odor and the smoke detectors’ ability to detect the smoke. Further, additional oxygen would have been provided to the smoldering fire. Although these would be appropriate steps to take in a situation involving fumes that might, for example, cause irritation or otherwise prevent the flight crew’s ability to operate the airplane, they are not appropriate in a situation in which a fire is suspected. Therefore, the Safety Board concludes that the increased airflow that resulted from the Fumes Evacuation checklist actions diluted the smoke and inhibited its detection by either the smoke detection system or flight crewmembers and provided the fire with additional oxygen. Although the FAA provides guidance to crewmembers on issues related to inflight fires in AC 120-80, the AC does not provide guidance to flight crews on how to respond to evidence of smoke, fire, and fumes in the absence of a cockpit alert. In 2004, an international initiative was undertaken to improve guidance and checklist procedures in this area. The initiative developed a smoke, fire, and fumes checklist template to standardize and optimize responses to such events.

[ANALYSIS] Unfortunately, the lack of information about the devices or the batteries prevented any determination of whether these batteries were associated with previously known recalls. The Safety Board concludes that the exact origin and cause of the in-flight fire on board the airplane could not be determined due to the destruction of potentially helpful evidence; however, the available evidence suggests that the fire most likely originated in container 12, 13, or 14. Analysis National Transportation Safety Board A I R C R A F T Accident Report 52 Smoke or Fire Detection System Certification Tests 2.5  Because the first smoke detector system activation did not occur until at least 20 minutes after the fire had initiated, the system did not perform in accordance with the performance standards established by the FAA (smoke detector system activation within 5 minutes of fire initiation). The Safety Board’s investigation revealed that the smoke detection system certification tests for the DC-8 were not conducted with cargo loaded on the airplane, nor were they required to be, and, therefore, the tests did not account for the effects of a loaded cargo area on smoke detection. The investigation also revealed that current smoke and fire detection system certification tests do not adequately test for “all approved operating configurations and conditions,” as required by 14 CFR 25.858. Although the accident airplane was certificated under CAR 4b, not Part 25, the Safety Board is concerned that current certification methods do not ensure that smoke and fire detection systems are operating in compliance with FAA regulations for smoke detection timeliness. AC 25-9A contains guidelines on how to conduct certification tests in compliance with FAA regulations and states that the tests should demonstrate that the system can detect a smoldering fire. However, although the AC proposes various acceptable smoke generators and materials to be used in the tests, it does not indicate whether the tests should be conducted with cargo loaded in the cargo compartment or whether cargo containers should be used. Information provided by FAA and Boeing personnel indicates that smoke and fire detection system certification tests are typically conducted with a smoke‑generating device in the open area of the compartment, which is more representative of a passenger airplane configuration than a cargo configuration with containers. They indicated that the tests were conducted without cargo containers because using an empty compartment results in greater smoke dilution, requiring the smoke detectors to be sensitive to a small amount of smoke. With cargo containers loaded in the cargo compartment, air exiting the air conditioning vents in the ceiling is primarily directed outward and downward toward the floor. The cargo containers also create a barrier that the smoke must traverse before it enters the open space of the cargo compartment where it can be detected by the smoke detection system. Neither of the effects of cargo containers—the ventilation changes and the smoke barrier—are accounted for during the certification tests; therefore, the tests do not ensure compliance with the performance criteria contained in 14 CFR 25.858, which requires that all approved operating configurations and conditions be tested.

[ANALYSIS] The cargo containers also create a barrier that the smoke must traverse before it enters the open space of the cargo compartment where it can be detected by the smoke detection system. Neither of the effects of cargo containers—the ventilation changes and the smoke barrier—are accounted for during the certification tests; therefore, the tests do not ensure compliance with the performance criteria contained in 14 CFR 25.858, which requires that all approved operating configurations and conditions be tested. The Safety Board concludes that the current certification test standards and guidance for smoke and fire detection systems on board many aircraft are not adequate because they do not account for the effects of cargo and cargo containers on airflow around the detection sensors and on the containment of smoke from a fire inside a container. The Board is concerned that a fire producing a small amount of smoke within a sealed cargo container may not be promptly detected by existing smoke detection systems. Therefore, the Safety Board believes that the FAA should ensure that the performance requirements for smoke and fire detection systems on cargo airplanes account for the effects of cargo Analysis National Transportation Safety Board A I R C R A F T Accident Report 53 containers on airflow around the detection sensors and on the containment of smoke from a fire inside a container and should establish standardized methods of demonstrating compliance with those requirements. Smoke Detection and Fire Suppression Systems on 2.6  Cargo Airplanes The accident airplane was not required to be equipped with a fire suppression system, and, as a result, the fire, which began as a smoldering fire in one of the cargo containers, was able to develop into a substantial fire that burned through the container and ceiling liner while the airplane was airborne. The Safety Board has had longstanding concerns about the lack of fire suppression systems in cargo compartments and has issued several safety recommendations in the last 20 years to address this issue. For example, as a result of the investigation into the May 11, 1996, accident involving ValuJet flight 592, the Safety Board issued Safety Recommendation A-97-56, which asked the FAA to expedite final rulemaking to require smoke detection and fire suppression systems for all class D cargo compartments. On February 12, 1998, the FAA issued a final rule that required the installation of fire suppression equipment in cargo compartments on board passenger aircraft; however, the requirement for a fire suppression system did not apply to cargo airplanes. In addition, as a result of its investigation of the September 5, 1996, fire involving FedEx flight 1406, the Safety Board issued Safety Recommendation A-98-78, which asked the FAA to require on-board fire extinguishing systems if they were deemed feasible. In response, the FAA indicated that current procedures regarding ventilation and depressurization were sufficient means to control a fire until the flight could land and that an on-board suppression system would add “considerable” weight to the airplane and reduce the amount of cargo that could be carried on board. At the UPS flight 1307 public hearing, the FAA indicated that it was considering an NPRM for the development of a new compartment classification, class F, which would require fire detection and suppression systems in which any means of chemical extinguishment and crew access could be used.

[ANALYSIS] Although the FAA provides guidance to crewmembers on issues related to inflight fires in AC 120-80, the AC does not provide guidance to flight crews on how to respond to evidence of smoke, fire, and fumes in the absence of a cockpit alert. In 2004, an international initiative was undertaken to improve guidance and checklist procedures in this area. The initiative developed a smoke, fire, and fumes checklist template to standardize and optimize responses to such events. Recognizing that time is critical in responding to any event, the proposed smoke, fire, and fumes checklist guidelines emphasize that flight crews should balance efforts to diagnose and resolve any fire hazard with efforts to evaluate the possibility of diverting; that a crew should have a simple, rapid, integrated checklist for responding to such a situation; and that the operator’s Smoke or Fumes Evacuation checklist should be performed only after the fire has been extinguished or at such time as smoke or fumes become the greatest threat. The guidelines further state that smoke removal may change the airflow and worsen the situation by fanning or masking an ignition source. Boeing is currently updating the flight and operations manuals for most of its product line to provide revised smoke, fire, and fumes checklists based on the template of the industry initiative. However, Boeing does not currently plan to revise the manuals of older airplanes like the DC-8, which will require individual operators of this airplane to work with the manufacturer and the FAA to develop such revisions for their own Analysis National Transportation Safety Board A I R C R A F T Accident Report 49 operations. Therefore, the need for FAA guidance remains especially important for older airplanes, including the accident airplane model. If UPS procedures had contained guidance similar to that in the 2004 smoke, fire, and fumes industry checklist, the flight crew would have had a simplified, appropriate procedure to address the problem. Therefore, the Safety Board concludes that the aviation industry initiative on smoke, fire, and fumes provides specific guidance on when and how flight crews should respond to evidence of a fire in the absence of a cockpit smoke and/or fire warning. The Safety Board believes that the FAA should provide clear guidance to operators of passenger and cargo aircraft operating under 14 CFR Parts 121, 135, and 91K on flight crew procedures for responding to evidence of a fire in the absence of a cockpit alert based on the guidance developed by the 2004 smoke, fire, and fumes industry initiative. Fire Detection and Growth 2.3  The first indication of the fire was the first officer’s query to the other crewmembers about the smell of burning wood, which occurred about 20 minutes before the main cargo compartment Cargo Smoke warning light activated, indicating that the fire had been burning for at least that long. However, the flight engineer saw no smoke either of the two times that he visually checked the main cargo compartment. This evidence indicates that the fire initially did not generate a significant amount of smoke and was most likely initiated as a smoldering fire inside a cargo container. The construction of the cargo containers, which results in restricted airflow in or out of the container, likely inhibited the growth and detection of the fire in its initial stages.

[ANALYSIS] At the UPS flight 1307 public hearing, the FAA indicated that it was considering an NPRM for the development of a new compartment classification, class F, which would require fire detection and suppression systems in which any means of chemical extinguishment and crew access could be used. Class E compartments would continue to be used in all-cargo operations. The FAA acknowledged at the hearing that recent technologies have made fire suppression systems on board cargo airplanes more feasible to operators and that it expects to receive input on new ideas and technologies for fire suppression systems on cargo airplanes as a result of the NPRM. FedEx stated at the public hearing that it has already developed an on-board cargo compartment fire extinguishing system, which testing showed completely extinguished a fire before it breached the container. The Safety Board commends the actions taken by FedEx to voluntarily develop a fire suppression system for its airplanes. The Safety Board concludes that the threat from cargo fires could be mitigated by the installation of fire suppression systems. Therefore, Analysis National Transportation Safety Board A I R C R A F T Accident Report 54 the Safety Board believes that the FAA should require that fire suppression systems be installed in the cargo compartments of all cargo airplanes operating under 14 CFR Part 121. As previously mentioned, FAA primary lithium battery flammability tests have concluded that Halon is not an effective means to suppress fires involving primary lithium batteries. Currently, the Safety Board is unaware of any fire suppression system that is effective on primary lithium battery fires. Therefore, although the installation of fire suppression systems in all cargo compartments on cargo-only aircraft, as recommended by the Board, would reduce the risks from a fire involving most cargo items, including secondary lithium batteries, this action would essentially have no effect on a primary lithium battery fire. Further, until such time that fire suppression systems are installed on cargo-only aircraft, secondary lithium batteries will continue to typically be transported in compartments without fire suppression systems. Therefore, the Safety Board concludes that flight crews on cargo-only aircraft remain at risk from in‑flight fires involving both primary and secondary lithium batteries. The Safety Board believes that PHMSA should require aircraft operators to implement measures to reduce the risk of primary lithium batteries becoming involved in fires on cargo-only aircraft, such as transporting such batteries in fire resistant containers and/or in restricted quantities at any single location on the aircraft. The Safety Board further believes that, until fire suppression systems are required on cargo-only aircraft, as asked for in Safety Recommendation A-07-99, PHMSA should require that cargo shipments of secondary lithium batteries, including those contained in or packed with equipment, be transported in crew-accessible locations where portable fire suppression systems can be used. Emergency Response Issues 2.7  Emergency Response Timeliness 2.7.1  After the flight had been cleared to land on runway 27R, the PHL ATCT local controller changed the landing runway to 27L because it was the runway designated for emergencies at PHL.

[CONCLUSIONS] United Parcel Service Company (UPS) guidance on hazardous materials information 18. retrieval and dissemination was inadequate, which resulted in UPS personnel not providing emergency responders with detailed information about the hazardous materials on board the airplane in a timely manner. The requirements of 49 19. Code of Federal Regulations 175.33(d) are not adequate because they do not require operators to provide hazardous materials information to emergency responders immediately upon notification of an accident. Testing and incident data indicate that lithium batteries can pose a fire hazard. 20. Because many incidents involving lithium batteries are exempt from reporting 21. requirements, the data regarding such incidents are incomplete, which has prevented a thorough assessment of the causes of these failures and the risks associated with transporting lithium batteries. An in-depth analysis of the causes of secondary and primary lithium battery failures 22. would improve the safe transportation of these batteries. The Pipeline and Hazardous Materials Safety Administration’s August 2007 final rule 23. regarding the transportation of lithium batteries did not establish sufficient levels Conclusions National Transportation Safety Board A I R C R A F T Accident Report 66 of safety for air transportation of small secondary lithium batteries (no more than 8 grams equivalent lithium content). Probable Cause 3.2  The National Transportation Safety Board determines that the probable cause of this accident was an in-flight cargo fire that initiated from an unknown source, which was most likely located within cargo container 12, 13, or 14. Contributing to the loss of the aircraft were the inadequate certification test requirements for smoke and fire detection systems and the lack of an on-board fire suppression system. National Transportation Safety Board A I R C R A F T Accident Report 67 Safety Recommendations 4. New Safety Recommendations 4.1  As a result of its investigation of the February 7, 2006, accident involving United Parcel Service Company flight 1307, the National Transportation Safety Board makes the following recommendations to the Federal Aviation Administration: Provide clear guidance to operators of passenger and cargo aircraft operating under 14 Code of Federal Regulations Parts 121, 135, and 91K on flight crew procedures for responding to evidence of a fire in the absence of a cockpit alert based on the guidance developed by the 2004 smoke, fire, and fumes industry initiative. (A‑07-97) Ensure that the performance requirements for smoke and fire detection systems on cargo airplanes account for the effects of cargo containers on airflow around the detection sensors and on the containment of smoke from a fire inside a container, and establish standardized methods of demonstrating compliance with those requirements. (A‑07‑98) Require that fire suppression systems be installed in the cargo compartments of all cargo airplanes operating under 14 Code of Federal Regulations Part 121. (A‑07‑99) Provide guidance to aircraft rescue and firefighting personnel on the best training methods to obtain and maintain proficiency with the high‑reach extendable turret with skin-penetrating nozzle. (A‑07-100) Require airport inspectors to ensure that Part 139 airports with cargo operations include cargo aircraft in their aircraft rescue and firefighting aircraft familiarization training programs. (A‑07-101) Require cargo operators to designate at least one floor level door  as a required emergency exit and equip the door with an evacuation slide, when appropriate. (A‑07-102) Require all emergency exits on cargo aircraft that are operable from the outside to have a 2-inch contrasting colored band outlining the exit. (A‑07‑103) As a result of this investigation, the National Transportation Safety Board makes the following recommendations to the Pipeline and Hazardous Materials Safety Administration:

[ANALYSIS] Analysis National Transportation Safety Board A I R C R A F T Accident Report 46 Analysis 2. General 2.1  The flight crewmembers were properly certificated and qualified under Federal regulations. No evidence indicated any preexisting medical or physical condition that might have adversely affected the flight crew’s performance during the accident flight. Although overnight flight schedules can be tiring and produce cumulative fatigue over successive nights,67 the accident occurred on only the second night of the trip sequence, and all of the flight crewmembers had made use of sleep opportunities during the day before the flight. Further, the accident occurred less than 2 hours into the trip, about 0000, a time at which neither pilot typically slept.68 Finally, there was no evidence of performance deficiencies that were clearly discernible and consistent with the known effects of fatigue. Therefore, the Safety Board concludes that no evidence was found indicating that fatigue degraded the performance of any of the flight crewmembers on the day of the accident. Examinations of the recovered components revealed no evidence of any preexisting powerplant, structural, or system failures. This analysis discusses the accident sequence, including the flight crew’s performance; the quality of fire and smoke emergency checklists; smoke and fire detection and suppression systems on cargo aircraft; cargo aircraft smoke and fire detection systems test certification requirements; ARFF training inadequacies; cargo airplane emergency exit requirements; hazardous materials information dissemination procedures; and issues related to the transport of lithium batteries on cargo airplanes. Flight Crew Performance 2.2  Actions During Descent and Landing Sequence 2.2.1  The accident flight was uneventful until just after the descent to PHL began, about 2335, at which time, the first officer detected an odor and asked the captain and flight engineer if they smelled anything. The captain stated during postaccident interviews that 67  P.H. Grander, K.B. Gregory, L.J. Connell, R.C. Graeber, D.L. Miller, and M.R. Rosekind, “Flight Crew Fatigue IV: Overnight Cargo Operations,” Aviation, Space, and Environmental Medicine, Vol. 69 (1998), pp. B26‑B36. 68  When off duty, the captain typically went to bed between 0100 and 0200 and the first officer between 0000 and 0100. Analysis National Transportation Safety Board A I R C R A F T Accident Report 47 he evaluated diversion alternatives at this time but decided to continue to PHL. The flight crew’s decision to continue to PHL is discussed further in section 2.2.2. After the first officer’s query, the flight crew began to actively analyze the situation and take action, including looking for visible evidence of smoke or fire in the area behind the cockpit. The flight crew also proactively began troubleshooting to determine the source of the odor. Neither the UPS DC‑8 AOM nor the DC-8 Airplane Flight Manual states what procedures should be accomplished in the event that only fumes are present.

[ANALYSIS] The flight crew also proactively began troubleshooting to determine the source of the odor. Neither the UPS DC‑8 AOM nor the DC-8 Airplane Flight Manual states what procedures should be accomplished in the event that only fumes are present. The flight crew’s use of UPS fire and smoke‑related checklists and industry efforts to upgrade such checklists are discussed in detail in section 2.2.3. The Cargo Smoke warning light in the main cargo compartment illuminated about 20  minutes after the first officer first mentioned the presence of an odor. One minute 15 seconds later, the Lower Cargo Fire warning light illuminated and indicated that smoke was in the lower aft middle cargo compartment. After the first warning light illuminated, the flight crew began to execute the Lower and/or Main Cargo Compartment Smoke or Fire checklist, and the first officer turned the airplane toward the airport. CVR evidence indicates that, during this portion of the flight, the first officer maintained control of the airplane and the captain oversaw the crew actions and worked with the flight engineer to handle troubleshooting, communication, and emergency efforts. Decision to Continue to Philadelphia International Airport 2.2.2  Investigators considered whether the flight crew should have executed a diversion to another airport when the odor was first detected. The flight crewmembers stated during postaccident interviews that they decided to continue to PHL because no smoke detector warning lights illuminated in the cockpit and no smoke was visually evident, which minimized their belief that an actual hazard existed. Further, the crewmembers also noted that unusual odors could be common from nonthreatening factors, such as flying over forest fires or from unusual cargo. Without supporting evidence, such as visible smoke and aural alerts, odor is an elusive and highly subjective factor for determining the presence of a hazard. Safety Board calculations indicated that, if the flight crew had begun a diversion at the first indication of an odor and performed a standard or emergency descent into Washington Dulles International Airport, Andrews Air Force Base, or Baltimore Washington International Airport, it would have taken about 17 to 18 minutes69 to land, compared to the 24 minutes it took to land at PHL. Because the flight was already established on the descent to PHL, a diversion would have added cockpit workload that might have affected the crew’s ability to focus its attention on the potential smoke or fire hazard. The Safety Board concludes that the flight crew’s continued descent to PHL was not inappropriate given that there was no evidence of abnormalities other than the odor and that no cockpit alerts had activated. 69  This time was based on the 12.5 minutes that calculations showed would have been required for the airplane to descend and reach any of the alternate airports and about 5 minutes for the flight crew to discuss and execute a diversion and to prepare and coordinate the approach with ATC.

[CONCLUSIONS > FINDINGS] An earlier problem with fumes in the passenger cabin contributed to the cockpit crew's failure to evaluate correctly the cause of the in-flight problem. 16. No in-flight emergency was declared. 17. No further action could have been taken by the cockpit crew to land the airplane more quickly after the in-flight problem was detected. 18. Because no in-flight emergency was declared and the airplane was not evacuated immediately after landing, passengers and crewmembers were unnecessarily exposed to the threat of fire for 1 1/2 minutes longer than necessary, and the flight attendants were not able to prepare the passengers before landing for a quick exit. 19. Emergency passenger evacuation procedures on passenger safety cards were not consistent with oral instructions. 20. Air carrier ground personnel increased the risk of spreading the cargo compartment fire by opening cargo compartment doors. 21. Air carrier ground personnel unnecessarily risked personal injury by entering the passenger cabin of the airplane to remove passenger belongings before the fire was extinguished and the airplane determined to be safe. 3.2 Probable Cause The National Transportation Safety Board determines the probable cause of the in-flight fire to be a chemical reaction resulting from a hydrogen peroxide solution, in concentration prohibited for air transportation, which leaked and combined with the sodium orthosilicate-based mixture from an undeclared and improperly prepared container. The probable cause of the unauthorized transportation was the shipper's lack of knowledge about restrictions and requirements for hazardous rnaterials and inadequate procedures for detecting undeclared hazardous materials shipments. Contributing to the delay in detecting the in-flight fire and the captain's decision not to declare an in-flight emergency was the lack of heat or smoke detection equipment in the cargo compartment and insufficient flightcrew communication. Contributing to the threat to the airworthiness of the airplane was the lack of a fire extinguishment system for the cargo compartment and the inadequate design of the cargo compartment ceiling.

[ANALYSIS] Because of the improved cargo liner flame penetration test requirements and the new restrictions limiting the size of class D compartments, Safety Recommendation A-81-13 was classified "Closed-- Acceptable Action” on August 11, 1986. The Safety Board urges the FAA to require fire detection and extinguishment systems in all class D cargo compartments; to review the certification of all types of cargo compartments to identify any aluminum o7 other components that fail to meet thermal protection requirements at least equal to cargo compartment liner thermal protection requirements; to consider the eifects of hazardous cargo involvement in fires in all types of cargo compartments; to require that safety deficiencies identified be corrected; and to inimediately evaluate prohibiting the transportation of oxidizers in ep Rete present class D cargo compartments, and determine if other classes of hazardous materials shou'd also be excluded frorn present class D cargo compartments. Adding these safaty systems to class D cargo compartments will provide even greater protection tian is presently provided by class C cargo compartments. 2.4 Operational Factors The review of the CVR and crew interviews indicates that a deficiency in communication occurred between the cockpit and cabin crews during the in-flight fire and the descent into Mashville. An examination of the dialogue among crewmembers suggests that the captain was skeptical about the flight attendant's initial report of smoke. The first officer also appears to have been reluctant initially to accept that smoke, rather than fumes, was in the airplane. Given the acknowledged seriousness of in-flight fire and the obvious association of 3 report of smoke in the cabin with a strong possibility of a fire, the Safety Board is deeply concerned by the Captain's apparent reluctance to accept either the flight attendant’s or deadheading crewmember’'s report as valid or to seek additional information to resolve his uncertainty. In order to understand the captain‘s reaction, the Board examined other circumstances that might have predisposed his behavior. Because the captain was aware of a mechanical discrepancy with the APU on an eartier flight which resulted in in-flight fumes, it would have been natural for tnis informatior. to influe..ce his perception of the initiat report of smoke. However, the A°U was Not operating; therefore, the captain should have dismissed it 2s being the source of any fumes. Further, with the flight only a few minutes away from landing, the captain was entering into a high activity level, and he had limited options availabte to deviate from the succession of events and activities already set in motion. That is, his current flight path, speed, and traffic sequence already was directed toward getting the airplane on the ground expeditiously, and he considered an expeditious landing the only immediate option available to alleviate this abnormal and ill-defined situation. The Safety Board believes that these circumstances may have operated in concert to precispose the captain to disbelieve the reports of smoke, and to establish a mind set that the cabin crew was instead experiencing the less serious fumes.

[ANALYSIS] Therefore, the Safety Board concludes that the cargo compartment failed to meet the intent of 14 CFR 25.857(d) and that the potential for a catastrophic in-flight fire existed. Because the cargo compariment was not equipped with fire or smoke detection systems, the cockpit crew had no way of detecting the threat to the safety of the airplane until smoke and fumes reached the passenger cabin. After smoke was detected in the passenger cabin, the cockpit crew had no means to identify the location of the fire. Furthermore, because the cargo compartment was not equipped with a fire extinguishment system, the cockpit crew had no means available to extinguish or suppress the fire in the cargo compartment. Without fire detection or suppression systenis, the cockpit crew must rely on the adequacy of cargo compartment designs and construction to control a fire in the cargo compartment. The Safety Board participated in the investigation of the accident involving a Saudia Lockheed L-1011 at Riyadh, Saudi Arabia, on August 19, 1980, in which an in-flight fire resulted in the deaths of all 301 passengers and crew aboard the airplane after it landed safely. The probable cause of the accident was determined to be an in-flight fire in a class D cargo compartment. Although the cargo compertment was equipped with an operative smoke detector device, the cargo compartment was not equipped with a fire extinguishment system. Asa result of its participation in the investigation, the Safety Board issued Safety Recommendations A-81-12 and -13 to the FAA on February 10, 198): A-81-12 Reevaluate the "Class D” certification of the L-1011 C-3 cargo compartment with a view toward either changing the classification to "C," requiring detection and extinguishing equipment, or changing the compartment liner material to ensure containment of a fire of the types likely in the compartment while in-flight. A-81-13 Review the certification of all baggagc/sargo compartments (over 500 cu. ft.) in the "D" classification to ensure that the intent of Title 14 Code of Federal Regulations 25.857(d) is met. In its recommendations to the FAA, the Safety Hoard noted several instances of fire in checked baggage from the ignition of matches and other items. fn most cases, the fires ignited while the aircraft were on the ground and the aircraft were not damaged. However, the Safety Board raised the possibility of such a fire while in flight and questioned the capability of class D compartments to contain a fire by “snuffing” it to keep it from spreading. In June 1983, the FAA Technical Center completed a project to study experimentatly the effectiveness of transport aircraft class D cargo compartments in containing fires through oxygen starvation.

[ANALYSIS] That is, his current flight path, speed, and traffic sequence already was directed toward getting the airplane on the ground expeditiously, and he considered an expeditious landing the only immediate option available to alleviate this abnormal and ill-defined situation. The Safety Board believes that these circumstances may have operated in concert to precispose the captain to disbelieve the reports of smoke, and to establish a mind set that the cabin crew was instead experiencing the less serious fumes. The captain’s skepticism about the report of smoke was also reflected in the first officer's dialogue with the cabin crew His cornments appear to be mere of a challenge of the accuracy of the reports than an effort to get additional details. Even atter Fe determined the problem in the cabin to be serious and after he recognized the need for timely firefighting assistance on landing, the first officer failed to aggressively recommend that crastVfire/rescue equipment meet the airplane. On identifying smoke in the passenger cabin, flight attendant No. 4 recognized the potential seriousness of the problern and without hesitation, even under “sterile cockpit” conditions, immedrately informed the first officer about the condition. Subsequent actions by the cabin crew, including efforts to locate the source of the fire, maintaining open communications with the cockpit, using a deadheading crewmember to evaluate and communicate information about the problem, and moving passengers from the affected area, also demonstrated that they considered the problem to be serious in conclusion, the Safety Board believes that while it is untikely that the captain could have taken any action to land the plane more quickly, the cockpit crew failed to use the cabin crew effectively to obtain an accurate understanding of the developing problem. Had communications between the cockpit crew and the cabin crew been more effective, the Safety Board believes that the captain would have called for fire/rescue equipment to meet the airplane and ordered an emergency evacuation on the runway. The Safety Board betieves that American Airlines should use this exarnple in cockpit and cabin crew coarcination training to illustrate the need for cockpit crews to more effectively use cabin crews in describing suspected in-flight safety problems and to emphasize the need for cabin crews to be assertive when communicating information about safety problems to cockpit crews. The Safety Board previously zddressed the issue of cockpit and cabin crew coordination training as a result of its investigation of the in-flight fire aboard a OC-9 at Cincinnati, Ohio, on June 2, 1983.?2 As a result of its investigation, the Safety Board issued Safety Recommendation A-84-76 which called for the FAA to require its principal operations inspectors to review air carrier training and if necessary, require amendments concerning actions flight crews should take for immediately and aggressively determining the source and severity of any reported cabin fire. In responses to this recommendation, on November 2, 19&4, anc March 7,1986, the FAA advised that it believed that current rules and quidance did not warrant further action.

[ANALYSIS] However, the Safety Board raised the possibility of such a fire while in flight and questioned the capability of class D compartments to contain a fire by “snuffing” it to keep it from spreading. In June 1983, the FAA Technical Center completed a project to study experimentatly the effectiveness of transport aircraft class D cargo compartments in containing fires through oxygen starvation. The study concluded that the Federal regulations did not ensure adequate burn-through resistance of class D cargo liners subjected to realistic fires. {{t noted that the cargo compartment liner is the initial fire barrier for the protection of aircraft components, structure, passenger, and crew from a fire inside the cargo compartment, and it noted that because of cabin exhaust ventilation airflow around the cargo compartment, an opening, rupture, or burn-through of any portion of the cargo liner could feed a cargo fire with large quantities of air. The report warned that some cargo compartments, although primarily lined with fiberglass, have aluminum components and that the use of aluminum may nullify the fire containment capability of burn-through resistant cargo compartment liners. Subsequently, on August 8, 1984, the FAA issued a notice of proposed rulemaking, Notice 84-11, that addressed the problem of fire containment in cargo compartments by specifying a new test method for determining the flame penetration resistance of compartment liners. When the Safety Board provided comments on the rulemaking on October 9, 1984, it advised the FAA that while proposed flame penetration test methods are more stringent than previous ones, a fire should not be allowed to persist ir: any state of intensity in an airplane without the knowledge of the flightcrew and that a fire detection system should be required for class D cargo compartments. On May 16, 1986, the FAA issued a fina! rule to amend fire safety standards for cargo or baggage compartments to become effective June 16, 1986. The final rule adopted more stringent cargo liner burn-through tests and smaller class D cargo compartments, but it rejected a requirement for fire detection systems in class D cargo compartments. furthermore, cargo compariment fire protection research and testing did not consider what effect hazardous materials involvement in a cargo fire could have on the capability of a cargo compartment to contain an in-flight fire. The FAA concluded in its final rufe that the effects of hazardous materials were beyond the scope of its rulemaking notice. However, the Safety Board believes that the incident aboard American 132 clearly demonstrates that hazardous materials involvement in a cargo compartment fire must be considered in all cargo compartment fire penetration safety standards; hazardous materials determined to present unacceptable threats should be prohibited. Safety Recommendation A-81-12 was classified “Closed--Acceptable Action” on November 2, 1982, following a commitment by U.S. air carriers to improve the fire containment capability of the cargo compartment by replacing Nomex fabric cargo compartment ceiting tiners with fiberglass. Because of the improved cargo liner flame penetration test requirements and the new restrictions limiting the size of class D compartments, Safety Recommendation A-81-13 was classified "Closed-- Acceptable Action” on August 11, 1986.

[CONCLUSIONS > FINDINGS] Ground personne! who are expected to respond to an aircraft when a fire is suspected shouid be trained on the appropriate actions to be taken. Further, airline personne! should be instructed not to board aircraft to collect the passengers’ carry-on luggage until the aircraft has been declared safe by fire personnel. 31 3. CONCLUSIONS 3.1 Findings 1. The flightcrew and the cabin crew were qualified and trained in accordance with FAA regulations and American Aistines requirements. 2. The airplane was maintained in accordance with applicable Federal regulations and company procedures. {{ 3. The hydrogen peroxide solution and sodium orthosilicate-based mixture were not properly q packaged, marked, labeled, or described for air transportation. 4. The hydrogen percxide solution was a 50 percent concentration solution and was prohibited in any quantity aboard the airplane. 5. The hydrogen peroxide solution teaked from the polyethylene drum before being loaded aboard American 132 and again in flight while aboard American 132. 6. A combination of the hydrogen peroxide sotution, sodium orthosilicate-based mixture, and the previously wet fiber drum caused the in-flight fire in the midcargo compartment. 7. The hazardous characteristics of the hydrogen peroxide solution and the sodium orthosilicate-based mixture and hazard warning information on containers of these materials at the shipper’s facilities should have been sufficient to have caused the shipper to have determined that they presented hazards and should have taken required precautions for shipping these materiats by air, irrespective of knowledge of specific Federal transportation regulatiors. 8. Notices to warn the public about restrictions governing the transportation of hazardous materials by air are inadequate. 9. Air carrier procedures for identifying undeclared hazardous materials in general freight are inadequate. 10. The cockpit crew had no positive means to identify the presence of an in-flight fire or its location because the cargo compartment had no fire or smoke detection system. 11. The class D midcargo cormpartment failed to meet the intent of 14 CFR 25.857(d) and resulted in the a. worthiness of the airplane being threatened when the fire breached the cargo compartrient. 12. The transportation of oxidizers in class O cargo cornpartments is unsafe because class O cargo compartments are intended to smother jires through oxygen starvation. 13. The aluminum support straps for the cargo compartment liners in the midcargo compartment were not protected sufficiently to provide thermal protection equivalent to the liners. 14, The cockpit crew erroneously concluded that furnes rather than smoke were present in the passenger cabin. 32 15. An earlier problem with fumes in the passenger cabin contributed to the cockpit crew's failure to evaluate correctly the cause of the in-flight problem. 16. No in-flight emergency was declared. 17. No further action could have been taken by the cockpit crew to land the airplane more quickly after the in-flight problem was detected. 18.

[ANALYSIS] In responses to this recommendation, on November 2, 19&4, anc March 7,1986, the FAA advised that it believed that current rules and quidance did not warrant further action. As a result, on May 12, 1986, the Safety Board classified Safety Recommendation A-84-76 “Ciosed--Unacceptable Action.” Subsequent to the Safety Board closing the recommendation, the FAA developed two proposed advisory circulars that addressed cabin safety training for crevemembers and improved coordination and communications among and between cockpit and cabin crews. The Safety Board commented in support of the FAA's proposals. The lack of close coordination and timely exchange of accurate information among crewmembers were clearly problems during preparations for a possible emergency landing of a DC-8 at Portland, Oregon, in 1978; during an in-flight fire aboard an t-1011 at Riyadh, Saudi Arabia, in 1980; during preparations for a possible ditching of an L-1011 near Miami, Florida, in 1935; and during an in-flight fire aboard a DC-9 at Cincinnati, Ohio, in 1985. These instances, as well as this in-flight fire, vividly support improved coordination and communications and joint cockpit and cabin crew training with respect to conducting emergency procedures and periodic emergency drills in which cockpit/cabin crew coordination and communication are practiced. 2.5 Evacuation and Survival Factors The lethal threat of smoke and fire in aircraft ta passenger safety and the need to remove passengers from that environment quickly is well acknowledged. Because the captain failed to order an emergency evacuation of the airplane until 2 minutes 8 seconds after touchdown, the passengers were unnecessarily exposed to these threats for about 1 1/2 minutes longer than necessary. The captain’s delayed decision also increased the time necessary to evacuate the airplane; therefore, flight attendants did not have time to use the public adcress system to prepare passengers for a quick exit or to provide clear, oral instructions to passengers on evacuation procedures. Consequently, while most passengers considered the evacuation orderly, some complained that they could not hear commands shouted by the flight attendants until they were near the exits. As a result, the evacuation was delayed when passengers were stopped at exits to remove their shoes and to discard their carry-on tuggage. The delayed decision to evacuat2 also prevented crashvfire/rescue personnel from being in place to assist in the evacuation and to protect passengers should the fire have broken through to the cabin. Arecraft Accident Report--Air Canada Fight 797 McDonnell Douglas OC-9-32, C-FTLU, Greater Cincinnats International Airport, June 2, 1983 (NISB/AAR-8409). The Safety Board concluded that the actions of the flight attendants were performed in accordance with American Airlines training and procedures.

[ANALYSIS] After takeoff, the captain was operating the controls of the airplane and the first officer was handling communications with air traffic controllers. Based on the CVR recording, the flightcrew’s performance was appropriate during the portions of the flight that preceded the crew’s first awareness that a problem existed. 2.4.2 Flightcrew Decisions and Actions During the Emergency Beginning at 1410:12, the flightcrew noted and verbalized concerns about electrical problems. At 1410:22, the captain stated, “We need, we need to go back to Miami.” This was followed 3 seconds later by shouts in the background of “fire, fire, fire.” Seven seconds later, as the first officer transmitted a request to ATC for clearance to return to Miami (and before receiving clearance from ATC), the airplane leveled off and began to descend. 122 49 CFR 175.81(a) states, “[p]ackages containing hazardous materials must be secured in an aircraft in a manner that will prevent any movement in flight which would result in damage to or change in the orientation of the packages.” 106 Based on the shouts from the passenger cabin recorded on the CVR cockpit area microphone at 1410:25 and the comment 2 seconds later, “we’re on fire, we’re on fire,” it should have been clear to both flightcrew members that a very serious emergency situation existed in the cabin. Although the captain decided immediately to return to Miami and initiated a descent, for the next 80 seconds the airplane continued on a northwesterly heading (away from the Miami airport) while the flightcrew accepted ATC vectors for a wide circle to the left and a gradual descent back toward Miami. The Safety Board evaluated the electrical system, engine, and flight control malfunctions that occurred in the 80 seconds during which the airplane continued northwestward away from MIA. The electrical problems that first made the flightcrew aware of the emergency (at 1410:12) likely were the result of insulation burning on wires in the area of the cargo compartment. Electrical system wiring is routed outside of the cargo compartment of the DC-9, in accordance with 14 CFR Part 25.869, which requires the wiring not to be located against the cargo compartment liner and to incorporate a high temperature insulation. Therefore, the flightcrew’s comments about the electrical problems indicate that the fire had probably already escaped the cargo compartment by 1410:12. (However, it probably had not yet burned through the cabin floorboards.) The flightcrew comments recorded by the CVR from 1410:12 through 1410:22 reflect the pilots’ concerns about and attention to these electrical problems. It is possible that these concerns continued to occupy some of the pilots’ attention during the initial period of their attempt to return to the ground. Another malfunction began at 1410:26, just as the shouts from the cabin would have alerted the flightcrew to the seriousness of the fire there. According to FDR data, while the left engine remained at its previous EPR setting, the right engine’s EPR decreased to the flight idle value.

[ANALYSIS] Given that the fire had progressed sufficiently to cause a tire in the forward cargo compartment to rupture at 1410:03, the investigation examined why there was not an earlier indication of smoke and/or fire in the cabin than the first audible report at 1410:25. Several factors might account for the lack of warning from smoke earlier in the fire sequence. First, the 120 The tests were not designed to be an exact replication or simulation of the circumstances of this accident, but were conducted to learn about the overall nature of a fire initiated by an oxygen generator and fed with high concentrations of oxygen released from additional oxygen generators. Because the investigation could not conclusively determine the exact physical arrangement of the generators in each box, the exact size of the boxes, the exact arrangement of the boxes and tires in the cargo compartment, or how many oxygen generators were initially activated, and because of differences between the test chamber and the accident cargo compartment, the Safety Board recognizes that the test results might differ somewhat from what occurred on the accident airplane. 121 See also the factors outlined in footnote 120, above. 103 cargo compartment liner is designed to limit the amount of ventilation to and from the cargo compartment; consequently, so long as the liner is intact, the smoke will not readily escape into the passenger compartment. Second, any smoke that did escape would not have readily entered the air flow in the passenger cabin, which comes from overhead and down into the area between the airplane outer skin and the cargo liner, then moves aft and exits through the outflow valve. Third, the oxygen generators would have initially fed the fire with an abundance of oxygen, tending to minimize the amount of smoke and resulting in a very rapidly developing fire. All of these factors in combination most likely prevented any noticeable migration of smoke from the forward cargo compartment into the passenger cabin or cockpit until relatively late in the development of the fire. Although black soot deposits on some of the overhead luggage compartments indicate that black smoke ultimately reached the passenger cabin, this smoke probably did not reach the passenger cabin until after the fire had breached the cargo compartment ceiling. Because the cargo compartment where the fire occurred was a class D cargo compartment and was not equipped (nor was it required to be equipped) with a smoke detection system, the cockpit crew of ValuJet flight 592 had no way of detecting the threat to the safety of the airplane from the in-flight fire until smoke and fumes reached the passenger cabin. Further, because the cargo compartment was not equipped (nor was it required to be equipped) with a fire suppression system, the cockpit crew had no means available to extinguish or even suppress the fire in the cargo compartment. If the fire started before takeoff, and a smoke/fire detection warning device had activated, the flightcrew most likely would not have taken off. However, the Safety Board concludes that even if the fire did not start until the airplane took off, a smoke/fire warning device would have more quickly alerted the pilots to the fire and would have allowed them more time to land the airplane.

[ANALYSIS] The flight attendants had completed ValuJet’s FAA-approved flight attendant training program. Visual meteorological conditions prevailed, and weather was not a factor in the accident. The evidence indicated that the accident airplane was equipped and maintained in accordance with Federal regulations and approved procedures. At the time of the accident, the three MEL [minimum equipment list] items and the one CDL [configuration deviations list] item had been deferred in accordance with approved lists; of these four items, the inoperative service interphone system will be discussed in section 2.4.3, and the remaining MEL/CDL items were not related to the accident sequence. There was no evidence of preexisting mechanical malfunctions or other discrepancies in the airplane structure, flight control systems, or powerplants that would have contributed to the accident. Based on the observed damage to rotating parts, both engines were developing power at ground impact. Evidence from the CVR revealed that about 6 minutes after takeoff from Miami, the crew of flight 592 became aware of a fire in the passenger cabin. Approximately 10 minutes after takeoff, flight 592 crashed into the Florida Everglades. The accident was not survivable. The catastrophic impact and destruction of the airplane precluded complete recovery of all airplane components. However, the wreckage that was recovered provided evidence of fire damage throughout the majority of the forward cargo compartment and areas of the airplane above it, with the most severe fire damage found in the ceiling area of the forward part of this compartment. Other areas of the airplane did not show significant fire damage, including the cockpit and the electronics compartment of the airplane located beneath the cockpit. The airplane’s electrical system was examined for indications as to what caused the electrical problems initially noted by the flightcrew. However, because so much of the wiring ran adjacent to the cargo compartment, and because so many of those wires were severely damaged, the source of those electrical anomalies could not be isolated. Examination of the heat-damaged wire bundles and cables revealed no physical evidence of short circuits or of burning that could have initiated the fire. Further, the heat and fire damage to the interior of the cargo compartment was more severe than the damage to the exterior, consistent with the fire having been initiated inside the cargo compartment. (The cargo compartment liner, which was designed to keep a fire contained within the cargo compartment, would also have functioned to keep an externally-initiated fire out of the compartment.) Finally, the heat-damaged wire bundles were not routed near the breached area of the cargo compartment, whereas the boxes containing the oxygen generators were loaded into the area directly beneath 101 the breached area of the cargo compartment. Thus, the electrical system was not a source of ignition of the fire. The investigation revealed that shortly before flight 592’s departure from Miami, five boxes of unexpended chemical oxygen generators and three tires (two of which included wheel assemblies)119 were loaded into the forward cargo compartment in the area where the fire damage was the most severe.

[ANALYSIS] Thus, the electrical system was not a source of ignition of the fire. The investigation revealed that shortly before flight 592’s departure from Miami, five boxes of unexpended chemical oxygen generators and three tires (two of which included wheel assemblies)119 were loaded into the forward cargo compartment in the area where the fire damage was the most severe. The investigation further found that safety caps were not installed over the percussion caps that start a chemical reaction in the oxygen generators; lanyards for the retaining pins for the percussion caps’ spring-loaded actuation mechanism were not secured on several generators; and the generators were not packaged adequately to prevent generators from striking the actuation mechanism or dislodging retaining pins on adjacent generators. Based on the results of the Safety Board’s fire tests on chemical oxygen generators that were conducted near Atlantic City, New Jersey, after the accident, the physical evidence of fire damage in the forward cargo compartment of the accident airplane, and the lack of other cargo capable of initiating a fire in the forward cargo hold, the Safety Board concludes that the activation of one or more chemical oxygen generators in the forward cargo compartment of the airplane initiated the fire on ValuJet flight 592. The Safety Board’s analysis, therefore, first examines the accident sequence, including the initiation and propagation of the onboard fire, and the adequacy of air carrier and FAA efforts to minimize the hazards posed by fires in cargo compartments of commercial airplanes. The analysis also explores the pilots’ performance and actions when they became aware of the fire shortly after takeoff from Miami, and the adequacy of smoke protection equipment and smoke evacuation procedures aboard air carrier aircraft. The analysis then examines the circumstances surrounding the shipment of the oxygen generators and the procedures for shipping company material and hazardous materials. The analysis also evaluates concerns raised regarding the adequacy of the FAA’s hazardous materials program; ValuJet’s outsourcing of maintenance and training activity; the company’s oversight of its contract maintenance facilities; and the FAA’s oversight of ValuJet and ValuJet’s contract maintenance facilities. Finally, the analysis addresses the adequacy of ValuJet’s procedures for manifesting lap children. 2.2 Propagation and Detection of Fire The first indication of a problem during the accident flight occurred at 1410:03, approximately 6 minutes after flight 592 took off from Miami, when the CVR recorded an unidentified sound, which prompted the captain to ask “What was that?” Simultaneously, an anomaly in the FDR altitude and airspeed parameters occurred consistent with a static pressure 119 Although the stock clerk and others who handled the tires believed that all three were mounted on wheel assemblies, evidence recovered at the accident site (including several pieces of oxygen generators and other debris found inside the unruptured main tires) indicated that one of the main gear tires was not mounted on a wheel assembly. 102 increase of about 69 psf. Within 12 seconds, the captain reported an electrical problem, and at 1410:25, there were voices shouting “fire, fire, fire” in the passenger cabin.

[ANALYSIS] If the fire started before takeoff, and a smoke/fire detection warning device had activated, the flightcrew most likely would not have taken off. However, the Safety Board concludes that even if the fire did not start until the airplane took off, a smoke/fire warning device would have more quickly alerted the pilots to the fire and would have allowed them more time to land the airplane. Further, the Safety Board concludes that if the plane had been equipped with a fire suppression system, it might have suppressed the spread of the fire (although the intensity of the fire might have been so great that a suppression system might not have been sufficient to fully extinguish the fire) and it would have delayed the spread of the fire, and in conjunction with an early warning, it would likely have provided time to land the airplane safely. Although class D cargo compartments are designed to suppress fire through oxygen starvation, this accident and events before this accident illustrate that some cargo, specifically oxidizers, can generate sufficient oxygen to support combustion in the reduced ventilation environment of a class D cargo compartment. The in-flight fire on American Airlines flight 132, a DC-9-83, on February 3, 1988 (see section 1.6.3.2), clearly illustrated the need for systems that would provide flightcrews with the means to detect and suppress fires in the cargo compartments of airplanes. As a result of its investigation of that accident, the Safety Board recommended that the FAA require fire/smoke detection and fire extinguishment systems for all class D cargo compartments (Safety Recommendations A-88-122 and -123). As recently as August 1993, although the FAA had investigated several incidents of fires that were initiated as a result of oxidizers in the cargo compartments of airplanes, the FAA responded to Safety Recommendations A-88-122 and -123 stating that fire/smoke detection and fire extinguishment systems were not cost beneficial, that it did not believe that these systems would provide a 104 significant degree of protection to occupants of airplanes, and that it had terminated its rulemaking action to require such systems. The Safety Board concludes that had the FAA required fire/smoke detection and fire extinguishment systems in class D cargo compartments, as the Safety Board recommended in 1988, ValuJet flight 592 would likely not have crashed. Therefore, the failure of the FAA to require such systems was causal to this accident. The crash of ValuJet flight 592 prompted the FAA to state in November 1996 that it would issue an NPRM by the end of the summer of 1997 to require, on about 2,800 older aircraft, the modification of all class D cargo compartments to class C compartments, which are required to have both smoke detection and fire extinguishment systems. The accident also prompted the ATA to announce in December 1996 that its members would voluntarily retrofit existing class D cargo compartments with smoke detectors. As of the date of this report, the Safety Board is unaware that any airplanes have been modified and are in service.

[ANALYSIS] ValuJet had established four emergency procedures for handling fire and smoke from electrical system and air conditioning (pressurization) system malfunctions, removing smoke from a pressurized airplane, and removing cockpit smoke from an unpressurized airplane. (See section 2.5 for a discussion of an alternate cabin smoke removal procedure developed by the airplane manufacturer.) Given the pilots’ clear awareness of smoke and fire aboard the airplane (based on their statements recorded on the CVR), the Safety Board evaluated the effect of the flight attendants’ actions on the flightcrew, the flightcrew’s use of the ValuJet smoke evacuation procedures and emergency equipment, and the adequacy of that equipment. When the flight attendant first opened the cockpit door at 1410:52, some smoke from the cabin area was likely introduced into the cockpit environment. However, during the 1 minute 42 seconds in which the CVR operated continuously after the emergency began (including the times that the cockpit door was open), the flightcrew made no comments about breathing or vision difficulties, nor were there any sounds of coughs from the crewmembers during this period.123 Based on the absence of comments and sounds indicating flightcrew physical impairment on the CVR, the Safety Board concludes that only a small amount of smoke entered the cockpit before the last recorded flightcrew verbalization at 1411:38, including the period when the cockpit door was open. However, the Safety Board is concerned that if the smoke concentrations on the cabin side of the door had been severe when the flight attendant opened the door, her actions could have resulted in the introduction of incapacitating smoke into the cockpit. In the event of a cabin fire, the cabin crew needs to immediately communicate information to the flightcrew, while maintaining a smoke barrier between the cockpit and cabin. The interphone would have been the most appropriate way to do this, but it was inoperative. Based on the FAA POI’s suggestion that an appropriate “alternate procedure” for an inoperative interphone might consist of a prearranged code for knocking on the cockpit door to gain entry, the Safety Board concludes that the current MEL requirements for the development of an “alternate procedure” for an inoperative service interphone are inadequate for a cabin fire situation.

[CONCLUSIONS > FINDINGS] The loss of control was most likely the result of flight control failure from the extreme heat and structural collapse; however, the Safety Board cannot rule out the possibility that the flightcrew was incapacitated by smoke or heat in the cockpit during the last 7 seconds of the flight. 13. Only a small amount of smoke entered the cockpit before the last recorded flightcrew verbalization at 1411:38, including the period when the cockpit door was open. 14. The current minimum equipment list requirements for the development of an “alternate procedure” for an inoperative service interphone are inadequate for a cabin fire situation. 15. There is inadequate guidance for air carrier pilots about the need to don oxygen masks and smoke goggles immediately in the event of a smoke emergency. 16. The pilots did not don (or delayed donning) their oxygen masks and smoke goggles, and in not donning this equipment, they were likely influenced by the absence of heavy smoke in the cockpit and the workload involved in donning the type of smoke goggles with which their airplane was equipped. 17. The smoke goggle equipment currently provided on most air carrier transport aircraft requires excessive time, effort, attention, and coordination by the flightcrew to don. 18. The sealed, plastic wrapping used to store smoke goggles in much of the air carrier industry poses a potential hazard to flight safety. 19. Emergency cockpit vision devices might have potential safety benefits in some circumstances. 20. Emerging technology, including research being conducted by the National Aeronautics and Space Administration, might result in improvements in the potential to provide passenger respiratory protection from toxic cabin atmospheres that result from in-flight and post-crash fires. 21. Because of the rapid propagation of the oxygen-fed fire and the resulting damage to the airplane’s control cables and structure, the use of the Douglas smoke evacuation procedures would likely not have affected the outcome. The Douglas DC-9 procedures involving partial opening of cabin doors for in-flight evacuation of smoke or fumes from the passenger cabin and similar procedures adopted by some operators of other transport-category airplanes might clear smoke sufficiently in the cabin (and prevent entry into the cockpit) to prolong the occupants’ survival time during some fire and smoke emergencies. 22. Given the potential hazard of transporting oxygen generators and because oxygen generators that have exceeded their service life are not reusable, they should be actuated before they are transported. 23. 24. 25. 26. 27. 28. 29. 30. 31. 135 Because work card 0069 did not require an inspector’s signoff at the completion of each task, and there was no requirement for it to do so, there might have been no inspection of the maintenance work related to the removal of the chemical oxygen generators. Had work card 0069 required an inspector’s signoff, one of the inspectors involved with the two airplanes might have noticed that safety caps had not been installed on any of the generators.

[ANALYSIS] Within 12 seconds, the captain reported an electrical problem, and at 1410:25, there were voices shouting “fire, fire, fire” in the passenger cabin. In the Safety Board’s fire tests,120 a main gear tire that had been inflated to 50 psi ruptured 16 minutes after the first oxygen generator was activated, when the fire destroyed 9 of the 12 tire sidewall plies. Because the tires in the accident airplane were loaded just forward of the cargo door, the tires would have been located just above the set of left static ports. The FDR altitude and speed data are based on readings from the left alternate static port, which is located on the left side of the fuselage at FS 341 between longerons 26 and 27, indicating that the unidentified sound on the CVR and the FDR anomaly at 1410:03 were most likely caused by the rupture of an inflated tire in the forward cargo compartment after the tire was partially burned through by the fire. Based on this sequence of events, the investigation analyzed when the fire on board the accident flight might have been initiated. Activation of a generator would have been most likely to occur during an event that could cause movement or jostling of the contents of the boxes. Accordingly, the Safety Board considered whether the fire might have been started as a result of one or more generators being activated during the loading process, which likely ended before 1340:29 when the passenger safety briefing was recorded on the CVR. The tire ruptured more than 30 minutes later. The Safety Board also considered whether the fire could have resulted from an oxygen generator being activated during the takeoff roll, which began about 1403:34. However, this was only 6 to 7 minutes before the tire ruptured. The Safety Board recognized that several factors could have affected the rate at which the accident fire developed, as compared to that seen in the Safety Board’s fire tests, including the relatively airtight design of the accident cargo compartment, the possible actuation of multiple oxygen generators, and the presence of more combustibles near the actuated generator than there were in the fire tests.121 Therefore, based on the general timing information obtained from the fire tests, the Safety Board concludes that one or more of the oxygen generators likely were actuated at some point after the loading process began, but possibly as late as during the airplane’s takeoff roll. Given that the fire had progressed sufficiently to cause a tire in the forward cargo compartment to rupture at 1410:03, the investigation examined why there was not an earlier indication of smoke and/or fire in the cabin than the first audible report at 1410:25. Several factors might account for the lack of warning from smoke earlier in the fire sequence.

[ANALYSIS] The task force is working on the manifest issues, including the subject of lap children. A report from the Secretary of Transportation is due to Congress by October 8, 1997. 133 3. CONCLUSIONS 3.1 Findings 1. The flightcrew was properly certificated and had received the appropriate training and offduty time prescribed by the Federal regulations. 2. There was no evidence that any preexisting medical condition affected the flightcrew’s performance. 3. The flight attendants had completed ValuJet’s Federal Aviation Administration-approved flight attendant training program. 4. Weather was not a factor in the accident. 5. The accident airplane was equipped and maintained in accordance with Federal regulations and approved procedures, and there was no evidence of preexisting mechanical malfunctions or other discrepancies in the airplane structure, flight control systems, or powerplants that would have contributed to the accident. 6. The activation of one or more chemical oxygen generators in the forward cargo compartment of the airplane initiated the fire on ValuJet flight 592. One or more of the oxygen generators likely were actuated at some point after the loading process began, but possibly as late as during the airplane’s takeoff roll. 7. Even if the fire did not start until the airplane took off, a smoke/fire warning device would have more quickly alerted the pilots to the fire and would have allowed them more time to land the airplane. 8. If the plane had been equipped with a fire suppression system, it might have suppressed the spread of the fire (although the intensity of the fire might have been so great that a suppression system might not have been sufficient to fully extinguish the fire) and it would have delayed the spread of the fire, and in conjunction with an early warning, it would likely have provided time to land the airplane safely. 9. Had the Federal Aviation Administration required fire/smoke detection and fire extinguishment systems in class D cargo compartments, as the Safety Board recommended in 1988, ValuJet flight 592 would likely not have crashed. 10. Given the information available, the ramp agents’ and flightcrew’s acceptance of the company materials shipment was not unreasonable. 11. ValuJet’s failure to secure the cargo was not unreasonable. 134 12. The loss of control was most likely the result of flight control failure from the extreme heat and structural collapse; however, the Safety Board cannot rule out the possibility that the flightcrew was incapacitated by smoke or heat in the cockpit during the last 7 seconds of the flight. 13. Only a small amount of smoke entered the cockpit before the last recorded flightcrew verbalization at 1411:38, including the period when the cockpit door was open. 14.

[CONCLUSIONS > FINDINGS] Because of the lack of information regarding products approved for transportation by the Bureau of Explosives, Research and Special Programs Administration cannot adequately ensure that these products are being packaged and shipped safely in the transportation environment. 46. ValuJet did not follow its internal procedures for boarding and accounting for lap children. 47. It is essential that air carriers maintain easily accessible and accurate records of the names of both ticketed and unticketed passengers aboard their flights for retrieval in the event of an accident or other emergency. 3.2 Probable Cause The National Transportation Safety Board determines that the probable causes of the accident, which resulted from a fire in the airplane’s class D cargo compartment that was initiated by the actuation of one or more oxygen generators being improperly carried as cargo, were (1) the failure of SabreTech to properly prepare, package, and identify unexpended chemical oxygen generators before presenting them to ValuJet for carriage; (2) the failure of ValuJet to properly oversee its contract maintenance program to ensure compliance with maintenance, maintenance training, and hazardous materials requirements and practices; and (3) the failure of the Federal Aviation Administration (FAA) to require smoke detection and fire suppression systems in class D cargo compartments. Contributing to the accident was the failure of the FAA to adequately monitor ValuJet’s heavy maintenance programs and responsibilities, including ValuJet’s oversight of its contractors, and SabreTech’s repair station certificate; the failure of the FAA to adequately respond to prior chemical oxygen generator fires with programs to address the potential hazards; and ValuJet’s failure to ensure that both ValuJet and contract maintenance facility employees were aware of the carrier’s “no-carry” hazardous materials policy and had received appropriate hazardous materials training. 138

[ANALYSIS] The interphone would have been the most appropriate way to do this, but it was inoperative. Based on the FAA POI’s suggestion that an appropriate “alternate procedure” for an inoperative interphone might consist of a prearranged code for knocking on the cockpit door to gain entry, the Safety Board concludes that the current MEL requirements for the development of an “alternate procedure” for an inoperative service interphone are inadequate for a cabin fire situation. Therefore, the Safety Board believes that the FAA should specify, in air carrier operations master MELs, that the cockpit-cabin portion of the service interphone system is required to be operating before an airplane can be dispatched. 123 The first officer coughed several times while the airplane was on the ground at Miami, but no coughing was recorded during the emergency portion of the flight. 109 Evidence recovered at the accident site indicates that the pilots were active in attempting to remove smoke from the cabin and cockpit before impact, and in doing so they had executed portions of the ValuJet emergency procedures for handling smoke. The soot pattern found on the outflow valve recovered in the wreckage is consistent with the flightcrew having at least partially opened the outflow valve using the manual method, which is part of the ValuJet electrical smoke/fire procedure for evacuating smoke. The four ValuJet emergency procedures for handling smoke and fire uniformly instructed the pilots to don their oxygen masks and smoke goggles, as the first item to be performed on the emergency checklist. However, the flightcrew comments recorded on the CVR sounded unmuffled. Further, these comments were recorded on the cockpit area microphone channel of the CVR; this microphone would not have picked up verbalizations made under an oxygen mask. This indicates that neither the captain nor the first officer donned their oxygen masks during the period of the emergency in which the CVR was operative and the pilots were speaking. The last recorded verbalization by the captain was at 1410:49; the last by the first officer was at 1411:38. Because smoke goggles of the type provided to the flightcrew must be donned subsequent to the oxygen mask to have any effect, the pilots probably did not don their smoke goggles from the onset of the emergency, at 1410:07, through at least 1411:38. There is no evidence to indicate whether they donned their masks and goggles after 1411:38. The donning of oxygen masks and smoke goggles at the first indication of smoke anywhere in the airplane can provide flightcrews with a sustained ability to breathe and see in the event of a subsequent influx of smoke into the cockpit. Although in this accident, the donning of oxygen masks and smoke goggles would not have assisted the crew in the initial stages of the emergency (because of the absence of heavy smoke in the cockpit), early donning of the smoke protection equipment might have helped later in the descent, if heavy smoke had entered the cockpit. Consequently, the Safety Board evaluated why the pilots of ValuJet flight 592 did not don their oxygen masks or smoke goggles while the emergency was in its early stages.

[ANALYSIS] The Safety Board is also aware that the National Aeronautics and Space Administration (NASA) has undertaken a new research program focused on mitigating the severity of survivable accidents. The Safety Board concludes that emerging technology, including research being conducted by NASA, might result in improvements in the potential to provide passenger respiratory protection from toxic cabin atmospheres that result from in-flight and post-crash fires. Therefore, the Safety Board believes that the FAA should evaluate and support appropriate 112 research, including the NASA research program, to develop technologies and methods for enhancing passenger respiratory protection from toxic atmospheres that result from in-flight and post-crash fires involving transport-category airplanes. 2.5 Emergency Procedures for Smoke Removal Although ValuJet adopted the DC-9 procedures developed by Douglas for clearing smoke from the cockpit (including, as a last resort for smoke originating in the cockpit, depressurizing the airplane and opening a cockpit side window to remove the smoke),125 ValuJet did not adopt a procedure developed by Douglas for the evacuation of smoke from the passenger cabin. This procedure calls for partially opening the right forward service door at the front of the cabin, then opening the passenger aft (tailcone) entrance door. According to Douglas, if these doors are opened, the “airflow will sweep smoke forward [to the open service door]” and the procedure is effective in clearing smoke from both the cabin and cockpit area. This procedure has been adopted by some operators of the DC-9, and similar procedures have been adopted by some operators of Boeing 747 airplanes, but the procedure has not been adopted by most U.S. carriers. The Douglas procedure was examined by the Safety Board in 1983 during its investigation of the Air Canada DC-9 in-flight fire accident. In that examination, the Board recognized the efficacy of the procedure in removing cabin smoke (based on flight test results provided by Douglas). Noting concerns expressed by some air carriers and fire protection experts (but not by the manufacturer, which disagreed) that the procedure could intensify a fire, the Safety Board stated that the outcome of using this procedure during the Air Canada accident sequence was highly uncertain. In this accident, the Safety Board concludes that because of the rapid propagation of the oxygen-fed fire and the resulting damage to the airplane’s control cables and structure, the use of the Douglas smoke evacuation procedures would likely not have affected the outcome. The Safety Board also recognizes that airlines that have not adopted these procedures might have what they believe to be legitimate safety reasons for that decision. Nevertheless, the Safety Board also concludes that the Douglas DC-9 procedures involving partial opening of cabin doors for in-flight evacuation of smoke or fumes from the passenger cabin and similar procedures adopted by some operators of other transport-category airplanes might clear smoke sufficiently in the cabin (and prevent entry into the cockpit) to prolong the occupants’ survival time during some fire and smoke emergencies.

[ANALYSIS] Consequently, the Safety Board evaluated why the pilots of ValuJet flight 592 did not don their oxygen masks or smoke goggles while the emergency was in its early stages. The training records of the captain and first officer substantiated that both pilots had received a single session of simulator training in the electrical fire and smoke emergency procedure during the ValuJet initial DC-9 qualification program (the first step of which was to don oxygen masks and smoke goggles). However, in the captain’s previous incident involving smoke in the cabin from an overheated air conditioning pack, she had obtained a successful outcome without donning the mask and goggles. This might have predisposed her to decide not to don an oxygen mask and smoke goggles when the emergency began on the accident flight. In an informal survey of air carriers conducted by the Safety Board, pilots from several air carriers indicated that they would not don oxygen masks and smoke goggles for situations such as reports of galley fire, smoke in the cabin, or a slight smell of smoke in the cockpit. Based on the circumstances of this accident and the results of its survey, the Safety Board concludes that there is inadequate guidance for air carrier pilots about the need to don oxygen masks and smoke goggles immediately in the event of a smoke emergency. The Safety Board believes that the FAA should issue guidance to air carrier pilots about the need to don 110 oxygen masks and smoke goggles at the first indication of a possible in-flight smoke or fire emergency. The Safety Board also considered that air carrier pilots, including those involved in this accident, might not don their smoke goggles immediately because the time and effort required to do so might interfere with their flight-related workload. The Board has no indication of problems with donning any of the current designs of quick-donning oxygen masks; however, based on its simulator evaluations of the time and effort required to don the smoke goggles provided to the accident flightcrew, the Safety Board recognizes that donning this equipment immediately could have detracted from crew coordination, aircraft control, and communications between the crewmembers and ATC. Therefore, as they attempted to return to Miami, the pilots had to balance the workload required to don this equipment against their requirements for crew coordination and airplane control. The Safety Board concludes that the pilots did not don (or delayed donning) their oxygen masks and smoke goggles, and that in not donning this equipment, they were likely influenced by the absence of heavy smoke in the cockpit and the workload involved in donning the type of smoke goggles with which their airplane was equipped.

[ANALYSIS] The accident also prompted the ATA to announce in December 1996 that its members would voluntarily retrofit existing class D cargo compartments with smoke detectors. As of the date of this report, the Safety Board is unaware that any airplanes have been modified and are in service. On June 13, 1997, the FAA issued an NPRM that would require the installation of smoke detection and fire suppression systems in class D cargo compartments. According to the NPRM, the airline industry would have 3 years from the time the rule became final to meet the new standards. The FAA indicated that it anticipated issuing a final rule by the end of 1997. The Safety Board is disappointed that more than 1 year after the ValuJet crash and 9 years after the American Airlines accident at Nashville, the class D cargo compartments of most passenger airplanes still do not have fire/smoke detection or suppression equipment and there is no requirement for such equipment. The FAA’s recent findings and the continued shipment of undeclared hazardous materials, including oxygen generators, highlight the importance of getting the proper equipment installed as rapidly as possible. Therefore, the Safety Board believes that the FAA should expedite final rulemaking to require smoke detection and fire suppression systems for all class D cargo compartments. 2.3 Acceptance of the COMAT Shipment The ValuJet lead ramp agent and flightcrew were trained to reject shipments marked with hazardous materials labels and be alert to the potential hazards of undeclared hazardous materials. However, because the five boxes of chemical oxygen generators that were delivered to the ValuJet ramp had no hazardous material markings or labels, and because the shipping ticket that the SabreTech employee provided to the ValuJet lead ramp agent indicated that the boxes contained empty oxygen canisters, neither the ValuJet lead ramp agent nor the ramp agent who loaded the boxes were provided with information to indicate the hazardous nature of the COMAT shipment. Based on the description of the COMAT on the shipping ticket, they might have assumed that the boxes contained empty (non-hazardous) oxygen cylinders. Therefore, the lead ramp agent likely was not prompted to discuss the contents of the COMAT shipment with the SabreTech employee or the flightcrew. The shipment also included three aircraft tires, the carriage of which was not prohibited by the hazardous materials regulations so long as the tires were not over-inflated. Although the lead ramp agent testified that he had shown the shipping ticket to the first officer, it is unlikely that the first officer would have considered empty oxygen canisters or aircraft tires as 105 potentially hazardous. Based on the available information, the flightcrew would have had no reason to know or suspect that hazardous materials were being proffered for carriage aboard the airplane.

[ANALYSIS] The Safety Board concludes that the sealed, plastic wrapping used to store smoke goggles in much of the air carrier industry poses a potential hazard to flight safety. Consequently, the Safety Board believes that the FAA should require that the smoke goggles currently approved for use by the flightcrews of transport-category aircraft be packaged in such a way that they can be easily opened by the flightcrew. The Safety Board is aware that emergency cockpit vision technology exists that might be applicable to improving flightcrews’ ability to see in the event of smoke in the cockpit. In this accident, based on the absence of flightcrew comments about smoke early in the sequence, the light sooting within the cockpit indicated by recovered wreckage, the likelihood that the flightcrew did not don smoke goggles (which need to be used with the emergency cockpit vision device), and the likelihood of severely degraded airplane controllability later in the sequence, the use of emergency cockpit vision technology would not have prevented this accident. Further, the Safety Board is concerned that flightcrews encountering a smoke emergency might devote valuable time and attention to rigging an emergency cockpit vision device, to the exclusion of the timely donning of their oxygen masks/smoke goggles and their execution of smoke removal procedures. However, the Safety Board concludes that emergency cockpit vision devices might have potential safety benefits in some circumstances. Therefore, the Safety Board believes that the FAA should evaluate the cockpit emergency vision technology and take action as appropriate. As a result of its investigation of the Air Canada DC-9 in-flight fire accident, the Safety Board recommended that the FAA “expedite the research at the Civil Aero Medical Institute necessary to develop the technology, equipment standards, and procedures to provide passengers with respiratory protection from toxic atmospheres during in-flight emergencies aboard transport category airplanes” (Safety Recommendation A-83-76). Based on the development of a joint international standard for passenger protective breathing equipment (PBE), the Safety Board classified this safety recommendation “Closed—Acceptable Action” on March 6, 1995. The Safety Board acknowledges that there are a variety of concerns about providing PBE to passengers (primarily based on the possibility that an emergency evacuation would be delayed while passengers don this equipment). Further, the Safety Board notes the emergence in recent years of potential alternative technologies for protecting passengers from a toxic cabin atmosphere caused by fires. The Safety Board is also aware that the National Aeronautics and Space Administration (NASA) has undertaken a new research program focused on mitigating the severity of survivable accidents. The Safety Board concludes that emerging technology, including research being conducted by NASA, might result in improvements in the potential to provide passenger respiratory protection from toxic cabin atmospheres that result from in-flight and post-crash fires.

[ANALYSIS] FAA tests showed that when this system was used to fight a fire, it delayed the onset of flashover, reduced cabin air temperatures, improved visibility, and increased potential survival time. The Safety Board is concerned about the number of losses that have occurred and concludes that currently, inadequate means exist for extinguishing on-board aircraft fires. Therefore, the Safety Board believes that the FAA should reexamine the feasibility of on-board airplane cabin interior fire extinguishing systems for airplanes operating under 14 CFR Part 121 and, if found feasible, require the use of such systems. The Safety Board realizes that requiring on-board extinguishing systems may not entirely resolve these safety concerns because they may become disabled by crash impacts. Further, the Safety Board realizes that the full implementation of such technology will require a number of years. Therefore, the Safety Board concludes that in addition to the safety benefits provided by on-board extinguishing systems, ARFF capabilities must also be improved so that firefighters are able to extinguish aircraft interior fires in a more timely and effective manner. Therefore, the Safety Board believes that the FAA should review the aircraft cabin interior firefighting policies, tactics, and procedures currently in use, and take action to develop and implement improvements in firefighter training and equipment to enable firefighters to extinguish aircraft interior fires more rapidly. 76 3. CONCLUSIONS 3.1 Findings 1. The flightcrew was properly certificated and qualified in accordance with the applicable regulations and company requirements. Evidence from crew duty time, flight time, rest time, and off-duty activity patterns did not indicate that behavioral or psychological factors related to fatigue affected the flightcrew on the day of the accident. 2. The smoke detection system installed on the airplane functioned as intended and provided the crewmembers with sufficient advance warning of the inflight fire to enable them to land the airplane safely. 3. The Boston Center air route traffic control center and New York terminal radar approach control controllers responded appropriately once they were aware of the emergency and provided appropriate and needed information to assist the crew in the emergency descent and landing. 4. The airplane was properly certificated, equipped, and maintained in accordance with applicable regulations. No evidence of systems, mechanical, or structural failures was found. 5. The flight engineer’s failure to pull the cabin air shutoff T-handle, as required by the “Cabin Cargo Smoke Light Illuminated” checklist, allowed the normal circulation of air to continue to enter the main cargo area, thereby providing the fire with a continuing source of oxygen and contributing to its rapid growth.

[ANALYSIS] Evidence from crew duty time, flight time, rest time, and off-duty activity patterns did not indicate that behavioral or psychological factors related to fatigue affected the flightcrew on the day of the accident. The smoke detection system installed on the airplane functioned as intended and provided the crewmembers with sufficient advance warning of the in-flight fire to enable them to land the airplane safely. The ATC personnel involved with the flight were all properly certificated and qualified. The Boston Center ARTCC and New York TRACON controllers responded appropriately once they were aware of the emergency and provided appropriate and needed information to assist the crew in the emergency descent and landing. The airplane was properly certificated, equipped, and maintained in accordance with applicable regulations. No evidence of systems, mechanical, or structural failures was found. 2.2 Flightcrew Performance 2.2.1 Crew Coordination Although the airplane was landed successfully, several required items were not accomplished during the descent and landing. The flight engineer failed to perform step No. 6 of the “Cabin Cargo Smoke Light Illuminated” checklist (pulling the cabin air shutoff T-handle).71 If he had done so, airflow would have been shut off to the main cargo deck area while being maintained to the cockpit. The Safety Board concludes that the flight engineer’s failure to pull the cabin air shutoff T-handle, as required by the “Cabin Cargo Smoke Light Illuminated” checklist, allowed the normal circulation of air to continue to enter the main cargo area, thereby providing the fire with a continuing source of oxygen and contributing to its rapid growth. However, the Safety Board could not determine the degree to which it might have contributed to the severity of the fire. The flight engineer also failed to complete step No. 7 of the “Cabin Cargo Smoke Light Illuminated” checklist (to maintain a 0.5 psi differential cabin pressure). As a result, the occupants were unable to immediately open and exit from the primary evacuation exits (the L1 71 Although the CVR recorded the flight engineer stating, “pull cabin air” at 0538:40, its position after the accident indicates that the cabin air shutoff T-handle had not been pulled. 58 and R1 doors) because the airplane was still pressurized. The flight engineer acknowledged that instead of manually maintaining the appropriate pressure differential, after he had placed the outflow valve control in the manual position, he only “cranked it open a couple of times [turns].” Because they were at 33,000 feet and operating on only one pressurization pack, the outflow valve would have been almost completely closed before the flight engineer cranked it. As demonstrated in the Safety Board’s test on a similar DC-10, manually cranking the outflow valve control two times will not perceptibly open the outflow valve from fully closed on a static airplane. The Safety Board concludes that the evacuation was delayed because the flightcrew failed to ensure that the airplane was properly depressurized.

[CONCLUSIONS > FINDINGS] No evidence of systems, mechanical, or structural failures was found. 5. The flight engineer’s failure to pull the cabin air shutoff T-handle, as required by the “Cabin Cargo Smoke Light Illuminated” checklist, allowed the normal circulation of air to continue to enter the main cargo area, thereby providing the fire with a continuing source of oxygen and contributing to its rapid growth. However, the Safety Board could not determine the degree to which it might have contributed to the severity of the fire. 6. The evacuation was delayed because the flightcrew failed to ensure that the airplane was properly depressurized. 7. The captain did not adequately manage his crew resources when he failed to call for checklists or to monitor and facilitate the accomplishment of required checklist items. 8. Crewmembers who do not use protective breathing equipment during a smoke or fire emergency may place themselves at unnecessary risk in attempting to address or escape from the situation. 9. Crewmembers may not be adequately aware that attempting to open a passenger exit door when the airplane is still pressurized may result in the door not opening. 77 10. The DNA synthesizer was not completely purged of volatile chemicals (including acetonitrile and tetrahydrofuran) before it was transported on board flight 1406. 11. The presence of the aerosol cans, the containers of acidic liquid, as well as several packages of marijuana on board the accident flight illustrates that common carriers can be unaware of the true content of many of the packages they carry. 12. The transportation of undeclared hazardous materials on airplanes remains a significant problem and more aggressive measures to address it are needed. 13. The Department of Transportation hazardous materials regulations do not adequately address the need for hazardous materials information on file at a carrier to be quickly retrievable in a format useful to emergency responders. 14. FedEx’s policy of providing information only to the Safety Board after the Safety Board initiates an investigation is inconsistent with the need to quickly provide emergency responders with essential information to assess the threat to themselves and the local community. 15. More effective preparation for emergencies involving hazardous materials and a system for coordination among the Air National Guard, Stewart International Airport management, and all local and State emergency response agencies are needed. 16. Airport emergency plans should specifically address hazardous materials emergencies. 17. Currently, inadequate means exist for extinguishing on-board aircraft fires. 18. In addition to the safety benefits provided by on-board extinguishing systems, aircraft rescue and firefighting capabilities must also be improved so that firefighters are able to extinguish aircraft interior fires in a more timely and effective manner. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was an in-flight cargo fire of undetermined origin. 78

[ANALYSIS] As discussed in section 1.6.4, when there is no electric power to the airplane the motor that operates the door is powered by a charged air bottle. If an attempt is made to open the door when the cabin pressure differential is above 0.5 psi, the bottle pressure will bleed off and the door will not open. Although the lack of the L-1 door as an escape route was not a significant factor in this accident, the Safety Board is concerned that under other circumstances the loss of a passenger exit door could have serious safety consequences. The Safety Board concludes that crewmembers may not be adequately aware that attempting to open a passenger exit door when the airplane is still pressurized may result in the door not opening. Therefore, the Safety Board believes that the FAA should require all Part 121 operators of airplanes that rely on air pressure to operate exit doors to make crewmembers aware of the circumstances of this accident and remind them of the need to ensure that the airplane is depressurized before attempting to open the passenger exit doors in an emergency. 2.3 Fire Initiation 2.3.1 Location from Which the Fire Might Have Initiated Because the fire burned for about 4 hours after smoke was first detected in the cabin cargo compartment, under changing conditions, much of the potentially helpful evidence was destroyed by the fire.76 The growth of the fire was likely affected by the failure to pull the cabin air shutoff T-handle and the opening of doors L1 and R1 about 0556, which (even though L1 did not open completely) would have provided additional ventilation (oxygen), and thus increased the rate of fire growth. An even greater addition of oxygen occurred about 0650 when the cargo door was opened. (Witnesses reported that flames first broke through the fuselage shortly after the cargo door was opened. The airport operations log recorded that flames breached the crown of the fuselage about 0655, which was 1 hour after the airplane landed and about 1 hour and 19 minutes after the illuminated smoke detector lights were first noted by the flightcrew.) Despite the length of time the fire burned and the resulting destruction of potentially 76 It should nonetheless be noted that even after the prolonged fire, many containers (including those in rows 1, 2, 3, 14, 15, and 16) were not severely compromised, or still contained substantial quantities of highly combustible materials (such as magazines, technical manuals, dry cleaning bags, bubble wrap, and clothing) that remained unburned. 62 helpful evidence, the postfire condition of the airplane and its contents were nonetheless examined for clues as to the location from which the fire might have initiated. One factor that investigators considered was the "V" burn pattern that originated at container 6R. It is a basic premise of fire science that such a “V” pattern often points to the origin of a fire.

[ANALYSIS] Although the investigation examined this as a possible ignition scenario, it could not be determined whether the chemicals in the synthesizer played any role in igniting the fire. The investigation could not develop a viable and convincing scenario to explain how the synthesizer could have started a fire. Further, although the cargo debris from all of the cargo containers that had been carrying general cargo was also examined for possible ignition sources, it was not initially examined in as much detail as the items known to have been loaded into 6R because of the volume and condition of the debris. When a more detailed examination of that cargo debris was conducted later in the investigation, no ignition sources were identified in that cargo debris. However, because of the deteriorated condition of the cargo debris and the possibility that some cargo had been completely destroyed in the fire, the Safety Board could not rule out that on the accident flight, an ignition source was present in one of those cargo containers. In light of the discovery of several shipments of marijuana on board the accident flight and the suggestion of one of the parties that marijuana is subject to spontaneous combustion, investigators considered this as a possible ignition source. The Safety Board recognizes that some organic matter, such as hay, can generate a biological/organic reaction producing heat and combustion if it is both wet and sufficiently compressed yet still exposed to oxygen. However, all of the marijuana shipments on board the airplane had been vacuum packed. (The police investigator who documented the marijuana seizures testified that the specially trained dog would have detected the presence of additional packages of marijuana, if any had been on board, even if the packages were completely consumed in the fire.) The police investigator who documented the marijuana seizures explained that shippers of contraband such as marijuana attempt to reduce the size of the package by “using a vacuum to vacuum out all the air and get it as compact as possible.” Thus, although the marijuana would have been compressed, there would have been little or no oxygen available to permit or support the biological reaction needed to lead to spontaneous combustion. Further, neither the police investigator nor any of the fire experts or consultants questioned during the course of the Safety Board’s investigation were aware of a fire being initiated by spontaneous combustion of a marijuana shipment. Therefore, spontaneous combustion of a marijuana shipment was ruled out as a possible ignition source. 66 Finally, all of the airplane systems were examined for possible ignition sources; the electrical system showed no evidence of arcing, and none of the other aircraft systems showed any evidence of malfunction. Further, neither the crew interviews nor the FDR or CVR data indicated any failures or malfunctions in the airplane’s systems that might have played a part in initiating the fire. Therefore, aircraft systems were ruled out as a possible ignition source for the fire.

[ANALYSIS] According to the Federal Register notice, the study was undertaken to compare the mission and requirements for civil airport fire services to those of the Department of Defense. On August 1, 1996, the Safety Board commented: [T]he current mission set forth in 14 CFR Part 139 to “provide an escape path from a burning airplane” no longer suffices. The Safety Board supports a full study of the mission statement by the FAA with a view towards providing adequate [ARFF] resources to rapidly extinguish aircraft interior fires and to extricate aircraft occupants from such interior fires. All aspects of this issue, including staffing, extinguishing agents, firefighter training, and response times, should be evaluated and compared with DOD standards to develop a broader mission statement that includes interior cabin fire suppression and extrication of aircraft occupants. 88 The assistant fire chief who served as the initial incident commander testified that at the flight engineer’s suggestion a telephone call was placed to the airplane manufacturer (Douglas) in an unsuccessful attempt to determine whether there were alternate means for entering the airplane. 89 Air Canada DC-9-32 in Covington, Kentucky, June 2, 1983 (23 persons killed by smoke/and or fire); USAir 737 collision with a Skywest Fairchild Metro 227 in Los Angeles, California, on February 1, 1991 (22 persons killed by smoke/and or fire); Northwest Airlines DC-9 collision with a Northwest 727 in Detroit, Michigan, on December 3, 1990 (eight persons killed by smoke and/or fire); Air Transport International DC-8-62 in Jamaica, New York, on March 12, 1991 (freight only); Ryan International Airlines B-727 in Hartford, Connecticut, on May 3, 1991 (freight only); and TWA Lockheed L1011 in Jamaica, New York, on July 30, 1992. 75 Accident history suggests that the environment inside a burning airplane’s interior may be beyond the current technological capabilities of fire departments to extinguish within adequate time frames to successfully evacuate occupants or protect cargo. The Safety Board is aware that the FAA has researched fire extinguishing systems for airplane interiors, including testing of a water spray system that would discharge water into a particular area of the airplane when triggered by sensors in that area. Because the system would discharge water only to a focused area of potential fire, it would minimize the total amount of water that would need to be carried on board, thereby reducing the weight penalty of such a system. FAA tests showed that when this system was used to fight a fire, it delayed the onset of flashover, reduced cabin air temperatures, improved visibility, and increased potential survival time. The Safety Board is concerned about the number of losses that have occurred and concludes that currently, inadequate means exist for extinguishing on-board aircraft fires.

[ANALYSIS] Another example is an in-flight fire that originated in the left rear lavatory of an Air Canada DC-9 that forced the flightcrew to make an emergency landing on June 2, 1983, at the Greater Cincinnati Airport. In that fire, the first evidence of fuselage breakthrough occurred at the front of the aircraft, significantly forward of the initiation point. In addition, in a fire that originated below the floor immediately aft of the right galley on a Delta 737 while it was on the ground at Salt Lake City, Utah, on October 14, 1989, the first fuselage breach was in the first-class cabin, well forward of the fire’s origin. 81 Although 6R also contained a small amount of unburned combustible material (for example, newspaper wrapped around the industrial valves loaded in the outboard aft corner of the container), unlike the contents of containers in rows 8 and 9, the vast majority of easily burned unprotected material in 6R was consumed by the fire. 64 that container sustained fire damage only to the forward outboard post. Similarly, containers 9R and 8R contained significant amounts of unburned combustibles (such as paper items) after the fire. In analyzing the significance of the unburned materials in rows 8 and 9, the Safety Board recognized that when material towards the top of the containers in those rows burned, it could have formed an insulating layer of charred debris that would have slowed the downward progress of the fire.82 Postaccident examination revealed a significant amount of burned material (as well as unburned material) in that area. However, the high-pressure water streams from the firefighting efforts would have disrupted the arrangement of the burned and unburned materials, thus erasing any obvious signs of an insulating layer. Finally, investigators also noted that there was a "V" soot pattern on the interior of the fuselage originating at the floor level of the junction between 8L and 9L, but this is consistent with the exposure of that area when the fuselage separated and the fire vented at that point and is probably not related to the location of the fire’s origin. Thus, in comparing the fire damage in 6R with that in rows 8 and 9, it is possible that the fire in those rows was as significant as that in the area of 6R, but it might have started at or near the top of a container and was unable to progress very far into the volume of cargo loaded into those containers. In sum, there was insufficient reliable evidence to reach a conclusion as to where the fire originated. 2.3.2 Ignition Source of the Fire Because of an initial recognition among the fire investigators participating in the investigation that “V” burn patterns are generally highly significant, investigators examined the contents of 6R to try to identify a possible source of the fire.

[ANALYSIS] One factor that investigators considered was the "V" burn pattern that originated at container 6R. It is a basic premise of fire science that such a “V” pattern often points to the origin of a fire. However, as explained in the National Fire Protection Association’s Guide for Fire and Explosion Investigations, NFPA 921, “each time another fuel package is ignited or the ventilation to the fire changes, the rate of energy production and heat distribution will change. Any burning item can produce a plume and, thus a ‘V’ pattern. Determining which pattern was produced at the point of origin by the first material ignited becomes more and more difficult as the size and duration of the fire increases.” [Par. 3-7]77 Several areas in the main cabin cargo compartment exhibited extensive fire damage; however, the deepest and most severe heat and fire damage was found in and around container 6R.78 More of 6R's structure was consumed than of any other container, and it was the only container that exhibited severe floor damage, which was likely caused by burning/melting Lexan wall material that fell inward onto the unprotected container floor. (Container 6R was one of the more sparsely loaded containers, enabling fire to reach the floor level without having to consume much cargo in the process.) Further, 6R was the only container to exhibit heat damage on its bottom surface, and the area below container 6R showed the most extensive evidence of scorching of the composite flooring material. In addition, the overall burn damage pattern to the cargo containers in the main cargo cabin showed that the deepest burned-out area centered over container 6R and that the cargo containers surrounding 6R and the contents in these containers were all burned to a greater depth along the sides common to container 6R. There was heat damage to the cabin floor just aft of container 9L, but there was no cargo container in this area on the accident flight. This damage to the floor was consistent with burning material falling from the burning fuselage crown or contents from cargo container 9L falling into this empty space. Two aerosol cans found in this area showed evidence that they had been heated by fire on the outside of the cans, and that they had overpressurized and ruptured. If the fire had not burned so long, the "V" burn damage pattern and the extensiveness of the fire damage to 6R would have been stronger evidence of a fire originating inside 6R. Further, the deep burn and severe damage found in container 6R could also be accounted for by the fact that it was relatively empty and therefore largely unprotected by cargo. Thus, the Lexan side walls and nylon curtain could have fallen directly onto the floor of 6R and burned there, resulting in the severe damage to the floor of 6R and the exterior surfaces of the synthesizer.

[ANALYSIS] As demonstrated in the Safety Board’s test on a similar DC-10, manually cranking the outflow valve control two times will not perceptibly open the outflow valve from fully closed on a static airplane. The Safety Board concludes that the evacuation was delayed because the flightcrew failed to ensure that the airplane was properly depressurized. The CVR transcript reveals that the flight engineer was overloaded and distracted from his attempts to accomplish the “Fire & Smoke” and “Cabin Cargo Smoke Light Illuminated” emergency checklists (in addition to his normal descent and before-landing checklist duties) by his repeatedly asking for the three-letter identifier for Stewart so that he could obtain runway data for that airport. After the accident, the captain said that he had allowed the first officer to continue flying the airplane during the emergency so that he could coordinate with ATC and work with the flight engineer on completing the checklists. This should have resulted in an effective apportionment of the workload among the three crewmembers, in that the flying pilot would not have been overly distracted from flying the airplane, the flight engineer would have received needed assistance with his duties,72 and the captain would have had the opportunity to oversee the actions of both. However, the Safety Board is concerned that, despite the captain’s stated intention to serve in a monitoring and coordinating role, he failed to provide sufficient oversight and assistance to ensure completion of all necessary tasks. The captain did not call for any checklists to address the smoke emergency, which was contrary to FedEx procedures.73 (The flight engineer initiated the "Fire & Smoke" and "Cabin Cargo Smoke Light Illuminated” checklists.) Nor did he explicitly assign specific duties to each of the crewmembers. The captain also did not recognize the flight engineer’s failure to accomplish required checklist items, provide the flight engineer with effective assistance, or intervene to adjust or prioritize his workload. In fact, the captain repeatedly interrupted the flight engineer during his attempts to complete the “Fire & Smoke” checklist,74 thereby distracting him further from those duties. 72 At the time of the accident, the flight engineer had only 188 hours as a DC-10 flight engineer and had been working for FedEx for less than 6 months. 73 The FedEx DC-10 Flight Manual indicates, under “Emergency and Abnormal Checklist Procedures,” that “Phase One [memory]” items are to be “performed when directed by the Captain.” Further, it states, “all checklists containing Phase One items should be requested by the Captain by name” and strongly recommends that the captain and flight engineer “work together on the review of the Phase One items and the accomplishment of the Phase Two items.” 74 At 0538:38 and 0539:13, the captain interrupted him to ask whether he had run a test on the smoke detector system, which is not an item listed on the checklist. 59 Further, the captain did not initiate the “Emergency Evacuation” checklist, which was required to be initiated during the preparation for landing.

[ANALYSIS] Container 9R’s aluminum roof and two of its three Lexan walls were consumed by fire, as was the upper portion of the third Lexan wall; its inboard cornerposts were intact, but its outboard forward and aft cornerposts measured 1 foot 3.5 inches and 1 foot 5 inches, respectively. However, 9L contained a significant quantity of undamaged materials81 with a low melting point (polyurethane, polystyrene, and polyethylene), and the cornerposts of 79 It should be noted that a FedEx mechanic who was standing near the cargo door indicated that at about the same time that flames were first seen venting from the aft section of the airplane in the 8/9 area, or possibly earlier, flames were also starting to break through the side of the fuselage in the area of 6L. His observations were not contradicted by the other eyewitnesses. Rather, they suggest that flames might have been breaking through the fuselage at more than one place at the same time. The video footage showing flames breaking through farther aft was taken from the right of the airplane and would not have captured a breakthrough of flames on the left side. 80 The breakthrough principle is illustrated by several previous main cabin aircraft fires. For example, a cabin fire occurred in Atlanta in 1995 on a ValuJet DC-9 after an engine failure started in the aft end of the cabin from a failed engine compressor disk that punctured a fuel line and penetrated the cabin. (See, National Transportation Safety Board. 1996. Uncontained Engine Failure/Fire, ValuJet Airlines Flight 597, DC-9-32, N908VJ, Atlanta, Georgia, June 8, 1995. NTSB/AAR-96/03. Washington, DC.) In a report submitted to the Safety Board, an FAA fire specialist noted that in that accident, in spite of rapid fire department response, the fire gutted the cabin and penetrated the fuselage skin just behind the cockpit—at the opposite end of the cabin from where the fire started. Another example is an in-flight fire that originated in the left rear lavatory of an Air Canada DC-9 that forced the flightcrew to make an emergency landing on June 2, 1983, at the Greater Cincinnati Airport. In that fire, the first evidence of fuselage breakthrough occurred at the front of the aircraft, significantly forward of the initiation point.

[ANALYSIS] Further, the deep burn and severe damage found in container 6R could also be accounted for by the fact that it was relatively empty and therefore largely unprotected by cargo. Thus, the Lexan side walls and nylon curtain could have fallen directly onto the floor of 6R and burned there, resulting in the severe damage to the floor of 6R and the exterior surfaces of the synthesizer. When Lexan is heated, it typically burns, melts, and puddles, producing heat that would be sufficient to cause the damage to container 6R and its contents. Thus, a fire that 77 See also Par. 3-7.2, which states, “[a]reas of great damage are indicators of a high heat release rate, ventilation effects, or long exposure. Such areas, however, are not always the point of fire origin.” 78 In every other cargo container there was a layer of unburned cargo covering the container floor. 63 originated outside of 6R but eventually spread to that area could have resulted in a similar damage pattern. Comments on the CVR suggest that the smoke detector activation sequence might have begun with detector number 9 and initially moved forward; this suggests that the fire might have started aft of row 6. Further, some of the first flames to have breached the crown were observed approximately above the area occupied by cargo container rows 8 and 9.79 Although the smoke detector activation sequence and location of the early breakthrough of flames cannot be considered reliable indicators of a fire’s initial location,80 a possible connection between these factors and the location of the fire’s origin could not be discounted. Therefore, the Safety Board also considered the possibility that the fire originated aft of container row 6. Although there was some significant burn damage to the containers in rows 8 and 9, their contents, and the surrounding area of the aircraft, this damage appeared to have been less than the damage in the area of container 6R. For example, container 8L’s aluminum roof and two of its three Lexan walls were consumed by fire, as was the upper two-thirds of the third Lexan wall; its inboard forward and aft cornerposts measured 5 feet 10 inches and 5 feet 2 inches, respectively, and its outboard forward and aft cornerposts measured 1 foot 4.5 inches and 3 feet 6.5 inches, respectively. Container 9R’s aluminum roof and two of its three Lexan walls were consumed by fire, as was the upper portion of the third Lexan wall; its inboard cornerposts were intact, but its outboard forward and aft cornerposts measured 1 foot 3.5 inches and 1 foot 5 inches, respectively.

[ANALYSIS] In sum, there was insufficient reliable evidence to reach a conclusion as to where the fire originated. 2.3.2 Ignition Source of the Fire Because of an initial recognition among the fire investigators participating in the investigation that “V” burn patterns are generally highly significant, investigators examined the contents of 6R to try to identify a possible source of the fire. All items found in, or known to have been shipped in, container 6R were examined in detail (see section 1.14.2.1.1). In particular, the Safety Board examined the DNA synthesizer as a potential source of ignition because of the chemical smell noticed inside the unit and because the other items in that container were ruled not likely to have provided a source of ignition. The nature and degree of the fire damage to the synthesizer, particularly the heavy damage to the internal circuitry of the synthesizer’s controller panel (which is made of a low-flammability material), was thought to be suggestive of a source of fuel inside the synthesizer. (However, given the presence of volatile chemicals, and possibly vapors, inside the synthesizer, it is possible that a fire ignited external to the synthesizer could have produced a damage pattern similar to that resulting from an internal ignition source.) 82 According to the National Fire Protection Association’s Guide for Fire and Explosion Investigations, NFPA 921, “[a] protected area results from an object preventing the products of combustion from depositing on the material that the object protects, or prevents the protected material from burning.” [Par. 4-15.2] 65 Tests of the liquids from the accident synthesizer showed that flammable chemicals (THF and acetonitrile) were still present in the bottles on the machine after the fire. The quantity of chemicals remaining in the synthesizer’s bottles after the fire was insufficient to have caused the extensive internal fire damage to the synthesizer and the cargo container. However, it is likely that significant amounts of the chemicals were consumed in the prolonged and intense fire and thus the synthesizer probably contained much larger quantities of these flammable chemicals before the fire. (The presence of firefighting agent inside most of the bottles and the damage to many of the tubes that entered the bottles indicates that the bottles had been open to the atmosphere during at least part of the fire sequence.) These volatile chemicals—particularly the THF—could ignite a fire. THF, which is highly flammable under any circumstances, can also form unstable peroxides that can explode on contact with certain other materials or autoignite (spontaneously explode) in sufficient concentrations. Although the investigation examined this as a possible ignition scenario, it could not be determined whether the chemicals in the synthesizer played any role in igniting the fire. The investigation could not develop a viable and convincing scenario to explain how the synthesizer could have started a fire.

[ANALYSIS] The “Emergency Evacuation” checklist includes depressurizing the airplane before landing. If this checklist had been initiated, it would have provided another opportunity for the crew to accomplish the necessary depressurization that was missed on the “Fire & Smoke” checklist. In addition, the captain told investigators that he did not initiate the emergency descent checklist, but said that he thought he had accomplished the items on that checklist by memory. Although the emergency descent checklist (titled “Rapid Depressurization/Emergency Descent”) was probably not applicable to this situation, the captain’s statement is troubling because it suggests a belief that checklist items can be adequately accomplished from memory alone. Finally, the CVR transcript indicates that the captain did not call for an emergency evacuation. (After the captain said “we need to get…out of here,” the flight engineer said “emergency ground egress.”) The Safety Board concludes that the captain did not adequately manage his crew resources when he failed to call for checklists or to monitor and facilitate the accomplishment of required checklist items. Therefore, the Safety Board believes that the FAA should require the principal operations inspector (POI) for FedEx to review the crew’s actions on the accident flight and evaluate those actions in the context of FedEx emergency procedures and training (including procedures and training in crew resource management) to determine whether any changes are required in FedEx procedures and training. 2.2.2 Crew’s Use of Emergency Equipment Within 48 seconds after the first indication of a problem, the crew donned oxygen masks, as required by the “Fire & Smoke” checklist. The captain elected not to don his smoke goggles because they did not fit over his eyeglasses and they were dirty and scratched. The first officer elected not to wear his smoke goggles because he felt that they unduly restricted his peripheral vision. The flight engineer put his smoke goggles on but subsequently removed them because there was no smoke in the cockpit. The Safety Board is concerned that cockpit smoke may affect crewmembers’ vision, imperiling their ability to operate the airplane or properly address the emergency. Evidence in this accident indicates that smoke did not enter the cockpit in significant amounts until after the crew landed and stopped the airplane. However, the Safety Board is concerned that under different circumstances, the failure of crewmembers to don smoke goggles or to keep the goggles on during an emergency could adversely affect the outcome. In connection with its investigation of the May 11, 1996, accident involving ValuJet flight 592,75 the Safety Board concluded that there is inadequate guidance for air carrier pilots about the need to don oxygen masks and smoke goggles immediately in the event of a smoke emergency.

[ANALYSIS] The Safety Board recognizes that after this accident Stewart revised its emergency plan, and that airport operations personnel at Stewart have acknowledged the need to address those deficiencies in the airport’s emergency plan. However, the Safety Board is concerned that FAA requirements87 do not specifically address the need to prepare for hazardous materials emergencies, and that other airports may be similarly unprepared for hazardous materials emergencies. The Safety Board concludes that airport emergency plans should specifically address hazardous materials emergencies. Therefore, the Safety Board believes that the FAA should require all certificated airports to coordinate with appropriate fire departments, and all State and local agencies that might become involved in responding to an aviation accident involving hazardous materials, to develop and implement a hazardous materials response plan for the airport that specifies the responsibility of each participating local, regional, and State agency, and addresses the dissemination of information about the hazardous materials involved. Such plans should take into consideration the types of hazardous materials incidents that could occur at the airport based on the potential types and sources of hazardous materials passing through the airport. The Safety Board also believes that the FAA should require airports to coordinate the scheduling of joint exercises to test these hazardous materials emergency plans. 86 The assistant fire chief on duty served as incident commander until 0700, when the fire chief arrived on scene. 87 14 CFR 139.325 specifies what must be included in airport emergency plans of airports certificated under Part 139. 74 Firefighters were positioned on scene before the airplane landed and began firefighting efforts immediately. Although the firefighters initially attempted to conduct an interior attack on the fire from the foyer area, the location of the cargo containers prevented them from approaching the seat of the fire. After the cargo door was opened, firefighters observed orange flames and heavy smoke in the airplane, and the incident commander evacuated them from the airplane. The initial incident commander’s decision to evacuate the firefighters from the interior of the airplane was appropriate given the danger posed by the smoke and fire-filled airplane. However, the initial incident commander acknowledged that use of the SPAAT tool to penetrate the fuselage was delayed while he attempted to accommodate the flight engineer’s request that damage to the airplane be minimized.88 Although it is not clear whether an earlier entry would have improved the effectiveness of the firefighting efforts in this case, the Safety Board is concerned that more aggressive measures to enter the airplane, such as use of a fuselage penetrating tool, were not taken sooner. The Safety Board notes that the ANG fire chief testified that based on “lessons learned” from this accident, if a similar situation were to occur, he would immediately “get right in there with a hand line and deploy some type of penetrating tool on the outer skin of the aircraft.” The Safety Board has long been concerned about the lack of success of airport fire departments in extinguishing interior fires.89 On June 4, 1996, the FAA published “Airport Rescue and Firefighting Mission Response Study,” in the Federal Register and invited comments from interested parties. According to the Federal Register notice, the study was undertaken to compare the mission and requirements for civil airport fire services to those of the Department of Defense.

[ANALYSIS] FAA regulations and company policies and procedures. The flightcrew was qualified and certificated properly and the flight attendants were qualified for the flight. Each flight and cabin crewmember had received the training and off-duty time prescribed by Canadian regulations. There was no evidence of any preexisting medical or psychological conditions that might have affected thfi! performance of the flight and cabin crews. Involved air traffic controllers were certificated properly, and each controller had received the training and off-duty time prescribed by FAA regulations. Accordingly, the Safety Board directed its investigation to the ignition and propagation of the fire; to ATC procedures; to the performance of the pilots and flight attendants after the fire was discovered; and to factors which affected the survivability of the passengers and crewmembers. 2.2 Fire Ignition.-The evidence substantiates a conclusion that when the smoke was detected by the flight attendants, there was a fire located within the vanity and/or the toilet shroud in the aft lavatory. Therefore, the Safety Board tried to identify all possible ignition sources in this area. Given the location of the fire at the time the smoke was discovered, the Safety Board identified five possible ignition sources: an incendiary or explosive device; deliberate ignition; a burning cigarette; the toilet flush motor; or the flush motor electrical harness. In addition to these five, the arcing damage found on the feeder cables of the left and right a.c. generators and the maintenance history of the airplane a.c. generating system led the Safety Board to investigate a sixth possible source of ignition-the generator feeder cables which were routed beneath the floor of the aft lavatory. Based on the examination of the physical evidence and the results of the FBI laboratory analysis, the Safety Board concluded that neither an explosive nor incendiary device was involved. Also, there was no evidence that the fire was deliberately set. . Since the tests of the materials used in the aft lavatory showed that they met the fire resistance criteria contained in 14 CFR 25, it would have been virtually impossible for either a lighted cigarette or sparks produced by electrical arcing to ignite the materials used in the construction of the lavatory. In order to ignite the lavatory partitions and walls, some combustible material capable of sustaining high temperature combustion for the amount time necessary to ignite the lavatory walls had to have burned. Therefore, in investigating the possibility that the fire was ignited by a burning cigarette, the Safety Board focused on two areas below the vanity which could have contained combustible materials and into which a cigarette might have fallen-the sink area containing the trash chute and receptacle and the adjoining- amenities section. Since the lavatory trash receptacle was the most logical place for combustible material to collect and since a burning cigarette could fall down the trash chute into the receptacle, the damage in this area was evaluated. Had a cigarette started a fire in the receptacle, the only propagation path out of the receptacle would have been the trash chute.

[PROBABLE CAUSE] AE. the pilot stopped the airplane, the airport fire department began firefighting operations. Flight attendants and passengers opened the left and rig:ht forward doors, the left forward overwing exit, and the right forward and aft overwing exits. About 60 to 90 seconds after the exits were opened, a flash fire engulfed the airplane interior. While 18 passengers and 3 flight attendants exited through the forward doors and slides and the three open overwing exits to evacuate the airplane, the captain and first officer exited through their respective cockpit sliding windows. However, 23 passengers were not able to get out of the plane and died in the fire. The airplane was destroyed. As a result of its investigation, the Safety Board determined that •.. the probable causes of the accident were a fire of undetermined origin, an underestimate of fire severity, and conflicting fire progress information provided to the captain. Contributing to the severity of the accident was the flightcrew's delayed decision to institute an emergency descent. APPENDIX H -2In its investigation, the Safety Board found sufficient evidence to substantiate the conclusion that 8. fire propagated through the amenities section of the aft lavatory and had burned undetected for almost 15 minutes before the smoke was first noticed. Although, the Board was not able to identify the origin of the fire, it was able to eliminate several possibilities and identify its general location. The evidence disclosed that the fire propagated through the lower part of the amenities section of the lavatory vanity. Since the direction of airflow below the vanity and above the toilet waste tank permitted venting of smoke, fumes, and hot gases away from the cabin and open area of the lavatory, the fire went undetected for almost 15 minutes until it began to penetrate the lavatory liner and rise behind and outboard the liner, subsequently penetrating the sidewall and ceiling seams of the lavatory liner. The Petitioner acknowledged that it has no new evidence to present with regard to the probable cause of the accident. However, the Petitioner presents three principal assertions for the Safety Board's review: first, the Safety Board did not properly assess the available evidence regarding the time factor in the captain's decision to descend and land the airplane; second, the use of the word "misconstrue" in finding No. 8 carries a connotation of error on the part of the captain; and third, the contributing statement of the probable cause suggests that the captain's decision to institute an emergency descent was delayed longer than a prudent captain would have waited under similar circumstances. With regard to the first issue, the Petitioner referenced page 61 of the Board's report, which stated, in part, that "The Safety Board believes that a precautionary emergency descent shpuld have been initiated as soon as it became evident that the fire had not been visually located and could not be attacked directly with extinguishant.

[PROBABLE CAUSE] But, that is not the first reason why I went back tq the cockpit; the first thing I say after ....• I can't go back itls too heavy, I think \'ie'd better go down" Exhibit 12A, page 10. But we will come back to that part later. The report says in the same paragraph: "Since the flight attendant in charge and the first officer were not able to determine the location of the fire they were not able to assess the severity of the fire. Consequently••• , they provided the captain with an inadequate assessment of the fire's severity.··. That is easier said than done, and I quote from the following reports. Passenger Arnold Friedman (Exhibit 6A): "The male FIA had sprayed it with an extinguisher. Heavy brown smoke billowed forward. The first officer went aft several times." Passenger Harry Mosely: ..••. the noise of the CO2 fire extinguisher when triggered. This sound came from the back of the aircraft. He looked back and sa\'I the smoke." Flight attt=!ndant J. Davidson: "She ingested some smoke which made her feel dizzy... Mr. Benetti was standing back while I was trying to walk through the smoke and said in his report: "The smoke had an electrical acrid smell and burned his throath.·' and Mr. Benetti managed only to discharge part of the C02. So the flight attendant and I were not able to determine the location of the fire because we COULD NOT, NOT PHYSICALLY POSSIBLE, TOO MUCH SMOKE. This is where the assessment of our story stops short of reality. The report continues in paragraph 4: "The Safety Board bel ieves that a precautionary emergency descent should have been initiated as soon as it became evident that the fire had not been visually located and could be attacked directly with extinguishant. This became known at 19:04:04 when the first officer came forward from his first inspection of the aft cabin area, about 3 minutes before the decision to begin an emergency descent•..• Again, Exhibit 12A, page 10, I say "1 canlt go back now, itls too heavy (smoke). I think we'd better go down:'. So that is exactly what the report says we should have done and for the same reasons except for the "precautionary emergency descent~. I had in mind a descent, not a fuel saving descent of course because of the nature itself of the reason why •• ./3 APPENDIX G we were going down, but not an emergency descent also as I testified because I did not know enough.

[PROBABLE CAUSE] Female flight attendant at 19:05:36: ··..• it seems to be subsiding now". Flight attendant in charge at 19:06:52: ..... it is almost cleared". That change of situation is confirmed by many passengers' statements in Exhibit 6A. Arnold Friedman: "There was a short interval in which the smoke subsided". Glen W. Davis: ··•••and the smoke cleared for a minute or two". In Exhibit 6A, page 39, the Human Factors Group Chairman's factual report: "Severa1 passengers reported a short interval, estimated by one passenger to be 5 minutes, during which the smoke subsided. ". I believe that is enough evidence to confirm that at the time "Ie \'/ere deciding for a descent, there was a significant change in our situation: a decrease and complete disappearance of smoke in the cabin. The report fails to recognize that fact in spite of all the reports in the exhibits. What is called "conflicting fire progress", "underestimate of fire severity", "miscontrued reports" ~/ere just a simple observation, a noting of the facts by the crew. Of course, we did not have the benefit of knowing the future and we believed what we saw, and so it was for everyone on board. I remember when I went back for the second time that some people had resumed their drinking and talking. There are a lot of questions that go through your mind in a moment like this: Was the C02 effective? Maybe, the motor pump blew, produced smoke and died? Remember the pump was very present in my mind. Those were reasonable assumptions at the time, much more than the extraordinary set of circumstances that the report says happened (page 54, paragraph 5): "The momentary smoke abatement noted by the first officer' and fl ight attendant in charge between 19:04:16 and 19:04:23 (right at the time of our decision to descend) was probably attributable to the dilution effect of opening the lavatory door and discharging the C02 into the area, reclosing the lavatory door, and the almost simultaneous failure of the lavatory vent line and the flush and fill pipe connections and check valve, all of which increased the ventilation rate beneath the toilet shroud and accelerated the flow of smoke and gases to the area below the lavatory floor and overboard..·• It is the timing and this amazing sequence of events that led us to believe that the situation had improved, and so it was for everybody on board, and no one that was not there can pretend otherwise. Again, all of this is very compatible with the lines of the report page 57, paragraph 4: .../5 APPENDIX G ··While an actual infl ight fire is an extremely rare occurence, all reports of smoke in the cabin must be regarded as potentially serious.

[PROBABLE CAUSE] In an emergency descent, I am required in my seat and f thought that I could do more, maybe, with my goggles and the fact that I vias caught by surprise by the acridity of the smoke the first time and I would not be this tiole. I wanted to give it another shot so that I could see or even better fight the fire. Again from the cockpit voice recorder, Exhibit l2A, at 19:04:46 the captain says: ··Okay go back \'/henever you can but don I t get yourse1 f incapacitated.'· and my answer is: "no problem, no problem··. Not knowing what we know now, was it that unrealistic to think that way? During all that time, I do have the motor pump in my head as a possible cause ••• Exhibit 12A, transcript of the cockpit voice recorder, at 19:02:50 just after the flight attendant's report of a fire and before I left the cockpit, I say: "Got all the breakers pulled:·. Again, just before I left, I say: "You got all the breakers pulled out?" and the captain answers: ··The breakers are all pulled, yeah:·. In the cabin, the smell seemed to be coming very possibly from something electrical according to passengers and flight attendants' reports in Exhibit 6A. So I associated the fire with the lavatory pump as I testified in Cincinnati, but I could not convince myself since the breakers were all out and therefore the pump was supposed to be dead. So this is what happened between 19:02:40 and 19:04:07 and I am convinced that Captain Cameron would have gone down as I suggested because we worked all month (May) together and we had a very good professional relationship. " BUT, and the next 3 minutes raised even more harsh critici~m from the N.T.S.B. report, we did not commence descent and that is because at 19:04:16 Mr. Benetti says (report page 3, paragraph 3): •••••didn·t have to worry, I think it's gonna be easing up:' and at 19:04:23, I say to the captain: ··It is starting to clear now.··. • •• /4 APPENDIX G I never said not to descend, but I realize it implied just that, and the captain looked back and saw the cabin clear. Female flight attendant at 19:05:36: ··..• it seems to be subsiding now". Flight attendant in charge at 19:06:52: ..... it is almost cleared". That change of situation is confirmed by many passengers' statements in Exhibit 6A.

[PROBABLE CAUSE] While the extent to which this time interval contributed to the severity of the accident could be debated, the Safety Board believes that the criticality of an early decision under such circumstances is not in issue. The facts known to the captain at 1904:07 should have been sufficient indicators of a potentially serious hazard that a precautionary descent should have been made immediately. The first officer for his part should have realized that the acrid smelling smoke probably was the result of an electrical fire, associated it with the tripping of the aft lavatory flush motor circuit breakers 11 minutes earlier, and informed the captain accordingly. Despite the fact the flight attendant in charge told the captain at 1906:42 that, "••• even though I could not see the source but it's definitely inside the lavatory," he, in fact, observed thick curls of black smoke emerging through the seams of the aft wall of the lavatory and believed that the fire was not in the trash container -and had also informed the first officer of his belief. His observations and assessment of the fire danger were inconsistent with his report to the captain. It is the Board;s belief that the: flight attendant in charge should have realized that his fire fighting effort was ineffective and that the temporary subsiding of the smoke was misleading. Therefore, the Board concludes that these crewmember reports influenced the captain's decision to delay the initiation of a descent. The delay increased the time for the fire to propagate and the time that passengers were exposed to the toxic environment before the airplane could be evacuated. With regard to the second issue, the Petitioner disagrees with finding No.8 on page 70 that the captain misconstrued the fire progress reports from his crew. The Petitioner believes that the captain relied on his crew's reports, and states that IlThere is a vast difference between believing an erroneous communication, and misinterpreting a communication." Further, the Petitioner contends that the evidence does not support finding No.8. Therefore, the Petitioner requests that the last sentence of finding No.8 be changed to read, "Due to persistent reports from various crewmembers that the fire was abating, he delayed his decision to declare an emergency and descend." After reviewing finding No.8, the Safety Board agrees that the word "misconstrued" may have conveyed the wrong intent. The Board agrees that the captain was misled by the reports from crewmembers that the smoke was SUbsiding. However, the Board continues to believe that, in the interest of safety, the captain should have been more assertive in seeking the nature of the fire instead of accepting the nonspecific comments of the crewmembers as assurance that the situation was under control. Had the crewmembers positively stated that the fire source had not been identified, the captain might have taken action sooner to descend in preparation for an emergency landing. With regard to the third issue, the Petitioner does not agree that contributing to the severity of the accident was the flightcrew's delayed decision to institute an emergency descent.

[PROBABLE CAUSE] Again, all of this is very compatible with the lines of the report page 57, paragraph 4: .../5 APPENDIX G ··While an actual infl ight fire is an extremely rare occurence, all reports of smoke in the cabin must be regarded as potentially serious. However, such reports often turn out to be smoke from an overheated flushing motor or waste ignited by a discarded cigarette in a trash receptacle designed to contain a fire, conditions which are normally identified and corrected by flight attendants without further consequences. Therefore, the Safety Board realized that there is a need to evaluate the situation before deciding on the emergency action required.··. So I did not just stand there but went back with my smoke goggles to evaluate the last events, touched the door and knew immediately that we had a major problem and there was nothing we could do except to get on the ground and fast. I told the flight attendants to prepare for emergency landing and at 19:07:11 told the captain (Exhibit 12A): • "1 don't like ,...hat is happening, 1 think ,...eld better go down, O.K:·. The captain concurred and so it was between 19:04:07 (first report to the captain) and 19:07:11 (second report) as the development of the situation caused the decision to descend to be delayed, again, not knowing what would happen 'after. So the N.T.S.B. report (page 57, paragraph 4) goes on to say: "As a result, about 5 minutes 30 seconds elapsed between the time that no. 2 flight attendant told the captain there was a fire in the aft lavatory and his decision to begin emergency descent ..• however in t~i~ case the time to make the decision appears excessive given the circumstances.... The decision to initiate emergency descent was in fact made at 19:07:11 (Exhibit 12A), which is not 5 minutes 30 seconds but 4 minutes 31 seconds. At 19:08:12, the first Mayday was heard by ATC, the one minute difference was the time that I took to take my seat, declare Mayday twice, select 7700 on transponder, select emergency powel~ "on·' and finally the third f4ayday was heard (Exhibit 2A Ouimet statement). The first two Maydays were not heard because of electrical failures. The left electrical side failed before I arrived, and as 1 was calling the first Mayday, the right side failed at 19:07:41 (Exhibit lOA, page 3/ Exhibit 12A, page 14 OFDR and CVR fai1ed/ N.T.S.B. report page 4 paragraph 2). Therefore there was no power to the "Emet"gency Bus" where no. 1 radio is supplied and consequently I was unable to transmit.

[CONCLUSIONS > FINDINGS] Crewmember reports that the fire was abating misled the captain about the fire severity and he delayed his decision to declare an emergency and descend. 9. Because of the delayed decision to descend, the airplane lost the opportunity to be landed at Louisville. Had the airplane been landed at Louisville, it could have been landed 3 to 5 minutes earlier than it actually did land at Cincinnati. The delayed decision to descend and land contributed to the severity of the accident. 10. A faulty ATC handoff did not delay significantly Flight 797's landing at Greater Cincinnati Airport. 11. The fire consumed the lavatory walls, propagated into the ceiling, and then began to move forward. Smoke, toxic fumes, and heated gases began to enter the cabin, spread forward, and collect along the ceiling of the cabin. 12. The flight attendants' passing out wet towels to the passengers and instructing them to breathe through the towels or through articles of clothing aided in the survival of some of the passengers. 13. The first officer turned off the air conditioning and pressurization packs in the belief that the airflow was feeding the fire. The resulting loss of circulation accelerated the accumulation of smoke, heat, and toxic gases in the cabin and likely decreased the time available for evacuation. 14. Three of the four overwing exit windows were opened by designated passengers who had been selected and briefed to open them by the flight attendants. 15. When the airplane stopped, smoke had filled the cabin and visibility within the cabin was almost nonexistent 2 to 3 feet above the cabin floor. -7116. A flashfire occurred within the cabin within 60 to 90 seconds after the doors and overwing window exits were opened. Flames from this fire were not evident until after the survivors had left the airplane. Flames from the original fire never were evident within the airplane or to persons on the ground. 17. This was a survivable accident. 3.2 Probable cause The National Transportation Safety Board determines that the probable causes of the accident were a fire of undetermined origin, an underestimate of fire severity, and misleading fire progress information provided to the captain. The time taken to evaluate the nature of the fire and to decide to initiate an emergency descent contributed to the severity of the accident. 4. RECOMMENDATIONS On July 11, 1973, the Safety Board participated in the investigation of the Varig Airlines Boeing 707 accident near Paris, France, in which 124 persons died after a fire erupted in the rear lavatory.

[PROBABLE CAUSE] However, in this case, the time to make the decision appears excessive under the circumstances. The Safety Board believes that a precautionary emergency descent should have been initiated as soon as it became evident that the fire had not been visually located and could be attacked directly with extinguishant. This became known at 1904:07 when the first officer came forward from his first inspection of the aft cabin area, about 3 minutes before the decision to begin an emergency descent." APPENDIX G Mr. James Burnett -1002 December 20, 1984 We agree wholeheartedly that the emergency descent should have been initiated as soon as it became evident to the captain that the fire had not been visually located ~nd could not be combated directly. However, the evidence presented in the report indicates that this did not occur at 1904:01 as stated in the Board Report. On page 61 of the report, the Board states, "such reports, (of smoke in the cabin), often turn out to be smoke from an overheated flushing motor or waste ignited by a discarded cigarette in a trash receptacle designed to contain a fire, conditions which are normally identified and corrected by flight attendants without further consequences." At 1906:52, the captain, by his own testimony, and as indicated on page 3 of the Board report, still believed the fire to be in the lavatory trash bin, and he did not decide to descend at this time because, "I expected it (the fire) to be put out." He had been led to this decision by a series of positive reports with respect to the firefighting efforts of the crew. At 1904:16, only 9 seconds after the first officer had suggested that they go down, the flight attendant advised the captain, "you don't have to worry I think it's gonna be easing up." Seven seconds later, at 1904:23, the first officer advised, "Okay, it's starting to clear now." At 1905:35, another flight attendant advised, "Captain, your first officer wanted me to tell you that Sergio has put a big discharge of CO in the washroom, it seems ~o be SUbsiding all right." Some time after 1909:01, the flight attendant who gave the first report advised the captain, "Getting much better, okay." At 1906:52, the flight attendant advised the captain, "C02, it was almost Jfalf a bottle and it now almost cleared." At 1906:54, the captain responded, "Okay, thank you." Between, 1904:01 and 1906:54, the captain received 5 separate communications from three different crewmembers indicating that the fire was being brought under control and that the smoke was subsiding.

[PROBABLE CAUSE] In this case, the captain first was told that there was a fire in the washroom, and he then was advised by his first officer that "it's too heavy, I think we'd better go down." Until that time, the captain did not have cause to believe that the fire would subside and he should have begun a descent immediately. Afterwards, he was misled by the crewmember reports. While the crewmembers should have realized that their comments could mislead and influence the captain's actions, the captain himself was ultimately responsible and he should have sought more positive verification that the fire had been extinguished. Further, the crewmember reports were only one element of information available to the captain. For example, the captain's comment "it's the motor" indicates that he did consider that the source of the fire might be electrical rather than paper burning in the wastebin. The captain should have known that an electrical fire which has progressed to the production of heavy smoke could be uncontained and out of control particularly when combined with his knowledge that circuit breakers had popped over 10 minutes earlier. This knowledge should have been sufficient to prompt the captain's immediate action. The Safety Board realizes that pilots must often make difficult, immediate decisions when confronted with abnormal or emergency situations. They often do not have the benefit of, or the time to weigh, all the facts. It is for this reason that flig'htcrews must be well trained to take the actions quickly that will be most effective in coping with the most severe and unexpected emergencies. The Safety Board had addressed its concerns in past safety recommendations regarding the need for effective cockpit resource management training to better prepare flight crewmembers to deal with complex, time-critical, decisionmaking situations, such as faced by this captain. Any indication of an in-flight fire justifies very rapid actions to land the airplane as quickly as possible. Therefore, the Board believes that the intent of the contributing factor statement must remain. However, the contributing factor statement has been revised since the final decision is the captain's alone. The Board considers the probable cause appropriate to emphasize the importance of an accurate and timely evaluation of the fire hazard by the flightcrew. ACCORDINGLY, -5APPENDIX H (a) The Petitioner's Petition for Reconsideration and modification of probable cause and findings of the airplane accident report on Air Canada Flight 797, McDonnell-Douglas DC-9-32, C-FTLU, Greater Cincinnati International Airport, Covington, Kentucky, June 2, 1983, is hereby granted in part. (b) The Board's original report is revised and a corrected report will be issued to the public which contains modifications to finding No.8 and the probable cause. (c) The probable cause is revised as follows: The National Transportation Safety Board determines that the probable causes of the accident were a fire of undetermined origin, an underestimate of fire severity, and misleading fire progress information provided to the captain.

[PROBABLE CAUSE] With regard to the first issue, the Petitioner referenced page 61 of the Board's report, which stated, in part, that "The Safety Board believes that a precautionary emergency descent shpuld have been initiated as soon as it became evident that the fire had not been visually located and could not be attacked directly with extinguishant. This became known at 1904:07 when the first officer came forward from his first inspection of the aft cabin area, about 3 minutes before the decision to begin an emergency descent." The Petitioner does not disagree with the Board's belief, but contends that the time at which this conclusion became apparent to the captain did not occur at 1904:07, and the Petitioner offers the following points from the Board's report to support its position: The Board acknowledged the occurrence of previous in-flight incidents involving lavatory fires which normally have been identified and corrected by flight attendants without further consequences; Trash receptacles in lavatories are designed to contain a fire; At 1906:52, the captain indicated that he believed the fire to be in the lavatory trash bin and that he did not decide to descend at this time because he expected the fire to be extinguished; The captain was misled by a series of reports from his crew indicating that the fire was under control and being extinguished. (These were five reports between 1904:07 and 1906:54 from three different crewmembers.) The captain sent the first officer aft to combat the fire 39 seconds after the time the Board believes it was evident to the captain that the fire could not be attacked directly with extinguishant; The captain would not have taken this action had he believed the fire was not accessible at that time. -3APPENDIX H The Petitioner therefore concludes that the captain did not become aware of the uncontrollable nature of the fire until 1907:11, when the first officer returned from the cabin and advised, III don't like what's happening, I think we better go down. Okay?" The descent was begun immediately after the first officer resumed his seat. All parties to the investigation were in agreement that the time factor involved contributed to the severity of the accident. The Board's rationale with regard to the delayed decision was that had the captain decided to descend immediately after being told by the first officer, at 1904:07, that, "... I can't go back now, it's too heavy, I think we'd better go down,1l the flightcrew could have conceivably landed the airplane 3 to 4 minutes earlier. While the extent to which this time interval contributed to the severity of the accident could be debated, the Safety Board believes that the criticality of an early decision under such circumstances is not in issue. The facts known to the captain at 1904:07 should have been sufficient indicators of a potentially serious hazard that a precautionary descent should have been made immediately.

[PROBABLE CAUSE] It is very unfortunate that the news media did not quote the Board's comments that the captain exhibited outstanding airmanship without which the airplane and everyone on board would certainly have perished, but blared in headlines from coast to coast and in at least two countries, "Crew's delayed deciision causes accident." The crew's reputation was therefore destroyed by conclusions reached by the Safety Board which were not in accordance with the evidence in the Board's possession. We therefore respectfully request that the Safety Board change the last paragraph of the probable cause to indicate that the delayed decision was based upon evidence which was presented to the captain upon 5 different occasions by 3 different crewmembers, which evidence indicated that the fire was abating and that the smoke was subsiding, and that an emergency descent would not be required. -103NOTES ON N.T.S.B. REPORT AIR CANADA DC-9 C-FTLU CINCINNATI, JUNE 2, 1983 Submitted by: FlO C. Ouimet APPENDIX G ··The N.T.S.B. determines that the probable causes of the accident were a fire of undetermined origin, an underestimate of fire severity, and conflicting fire progress information provided to the captain. Contributing to severity of the accident was the flight crew's delayed decision to institute emergency descent.- So the criticism of the crew is contained in a time frame between 19:02:40 and 19:08:12 E.D.T., or the first 5 minutes and 32 seconds. Most of the explanation of that criticism is found on page 57 of the report. Paragraph 2, half way says: ··Neither he (flight attendant in charge) nor the first officer told the captain that they had not seen the fire and that they did not know exactly where it was or how lntensely lt was burning.··. J The board must have missed what is said by Mr. Benetti. on page 13 of Exhibit 12A, which is the transcript of the cockpit voice recorder. ··1 was able to discharge half of the COZ inside the washroom even though I could not see the source but it's definitely lnside the lavatory"· Those are the exact words of Mr. Benetti to captain Cameron. The report continues with paragraph 3: ··He (first officer) retreated the first time because he did not have smoke goggles with him...... In Exhibit 2A, first officer statement ··1 try to get tq the washroom but could not because of smoke··.· In the same N.T.S.B. report on page 3, paragraph 2 it says "The first officer said that he could not get to the aft lavatory because the smoke, which had migrated over the last three to four rows of seats, was too thick··. Also in Exhibit 12A, transcript of the voice recorder page 10, I say to the captain during my first visit to the cockpit ...•• 1 can't go back now, it's too heavy··. So I did not retreat and never said I did because I did not have my·smoke goggles, but because I could not physically walk to the washroom in that smoke. 1 thought •• •/2 APPENDIX G that, maybe, maybe using the smoke goggles would help me to get there with ~y tie around my nose and mouth.

[PROBABLE CAUSE] Had the crewmembers positively stated that the fire source had not been identified, the captain might have taken action sooner to descend in preparation for an emergency landing. With regard to the third issue, the Petitioner does not agree that contributing to the severity of the accident was the flightcrew's delayed decision to institute an emergency descent. While the Petitioner agrees that the time delay involved in commencing a descent contributed to the severity of the accident, the Petitioner believes that the cbntributing factor statement incorrectly implies that the captain's decision was different APPENDIX H -4from the decision that a more prudent captain would have made under similar circumstances. The Petitioner maintains that the captain did not have the benefit of the amount of time the Safety Board had to analyze the situation and make a decision. The Petitioner states, "... it is grossly unfair and unrealistic for the Safety Board to expect a flight crew to evaluate an incident in a dynamic situation, and reach the same conclusions in a few seconds that it took the Safety Board and all of its staff over a year to reach." Additionally, the Petitioner states that the Board could not even agree on the time when the crew's decision to descend should have been made during its "Sunshine" meeting on July 20, 1984. Therefore, the Petitioner requests that the contributing factor statement be changed to indicate that the delayed decision to descend was based on evidence provided by three different crewmembers on five different occasions which indicated that the smoke was abating and that an emergency descent would not be required. While it may appear to the Petitioner that the Safety Board's report unjustly focused on the flightcrew's delayed decision to land the airplane, the Board's conclusion was not made without a careful and lengthy review of the facts. The Board did come to an agreement regarding the time the flightcrew should have made a decision to descend1904:07-which is evident in the report, and there were no dissenting opinions filed by the Board's members after the "Sunshine" meeting. The Safety Board is of the opinion that the Petitioner's suggested change would not agree with the basic rationale of the Board's analysis of the flightcrew's handling of the fire situation. The Board believes that all pilots should be well a~are of the potential catastrophic nature of a cabin fire during flight. Therefore, the Board believes that any indication of a fire should prompt a flightcrew to begin immediate preparation for an expedient landing. In this case, the captain first was told that there was a fire in the washroom, and he then was advised by his first officer that "it's too heavy, I think we'd better go down." Until that time, the captain did not have cause to believe that the fire would subside and he should have begun a descent immediately. Afterwards, he was misled by the crewmember reports.

[PROBABLE CAUSE] It should also be noted that at 1904:46, 39 seconds after the Board indicates that it was evident to the captain that the fire could not be attacked directly with extinguishant, the captain sent the first officer aft to combat the fire. He would not have done so had he believed that the fire was not accessible at that time. Therefore, the above evidence indicates that the first time it became evident to the captain that the fire was not controllable was at 1901:11, when the first officer returned from the cabin and advised, "I don't like what's happening, I think we better go down, Okay?" Upon being informed for the first time that the fire was not under control, the captain made the decision to begin the descent immediately. Descent was begun as soon as the first officer resumed his seat and the aircraft prepared for an emergency descent. On page 10 of the Board report, under Finding number 8, the Board states, "Thereafter, he misconstrued reports that the fire was abating and he delayed ~i~ decision to declare an emergency and descend." Webster's dictionary efines the word "misconstrue" to mean "to misinterpret; misunderstand". The captain did not misinterpret, nor did he misunderstand any of the communication from his crew. He simply relied upon his crew and believed what he was told. There is a vast difference between believing an erroneous communication, and misinterpreting a communication. The word "misconstrue" carries a connotation of error on the part of the captain. It indicates that he misinterpreted what he was told, and thereby committed an error. There is simply no evidence to support such a conclusion. To believe is not to misconstrue. Mr. James Burnett -1013 APPENDIX G December 20, 1984 The Safety Board's report contains many more erroneous conclusions based upon misconstrued evidence. However, these errors are succinctly described in First Officer C. Ouimet's notes on the N.T.S.B.'s report. The notes have been provided.to the Safety Board, and we will not reiterate them here. We will state, however, that we fully support Mr. Ouimet's conclusions.

[ANALYSIS] Nonetheless, on final impact with the building, both of them were thrown forward into the galley bulkhead, action that could have incapacitated them. Except for minor contusions, both of them were able to respond and facilitate the evacuation from the R-2 exit. Although releasing their restraints was intended to speed up the evacuation, the possible consequences of serious injury could have prevented either or both of them from assisting in the evacuation. The Safety Board believes that the potential for flight attendant survival can be significantly increased by providing flight attendants with supplemental training to underscore the importance of remaining in their jump seats with their restraints fastened until the airplane has come to a complete stop. 2.6.2 Source and Migration of the Cabin Fire When the B-737 overrode the Metroliner, the cockpit and forward lower cargo bay areas were extensively damaged. As the B-737 and Metroliner continued to slide, the fuselage and lower cargo bay of the B-737 were involved with fuel from the Metroliner’s ruptured fuel cells and hydraulic fluid from the B-737’s damaged nose gear. The initial impact with the Metroliner also damaged the avionics bay located below the cockpit in front of the lower forward cargo bay. The front portion of the cargo bay collapsed rearward and upward. The location of the crew oxygen cylinder on the forward right side of the cargo compartment shows fuselage skin penetrations originating from outside of the airplane. The regulator for the crew oxygen cylinder was most probably damaged during the initial impact sequence which resulted in the escape of gaseous oxygen. Fuel from the Metroliner and hydraulic fluid from the B-737 provided a fuel source for the fire, and oxygen from the crew oxygen cylinder accelerated it. After the initial impact, the R-1 flight attendant, who was seated in the jump seat located directly above the cargo bay, remembered hearing 67 metal scrape just before the cabin lights went out and the emergency lights came on. He remembered the floor directly in front of his jump seat moving up and down about knee high as heat and smoke entered the forward cabin area. When the B-737 impacted the abandoned fire station and the airplane stopped, he recalled that the smoke coming through the floor near him became more dense and that it became more difficult to breath. He also noted that the first-class cabin filled with smoke very quickly. The significant fire damage in the forward cargo bay and the vertical burnthrough in the forward cabin area strongly suggest that the area was subjected to prolonged exposure to a high-temperature fire. That factor, as well as the relatively uniform burn pattern throughout the cabin and the fact that the B-737’s fuel did not contribute to the fire, indicates that the origin of the fire was in the forward cargo bay area. The extent to which the release of oxygen from the crew emergency cylinder accelerated the fire is unknown. However, assuming fuel from the Metroliner had penetrated the lower cargo area, oxygen released from the bottle would have enriched the burn environment and thereby accelerated the generation of heat and smoke.

[ANALYSIS] The extent to which the release of oxygen from the crew emergency cylinder accelerated the fire is unknown. However, assuming fuel from the Metroliner had penetrated the lower cargo area, oxygen released from the bottle would have enriched the burn environment and thereby accelerated the generation of heat and smoke. The presence of a melted and burned through area on the right outboard side of the fuselage, approximately where the crew oxygen bottle was secured to the right sidewall, is a further indication that a gaseous oxygen release served to accelerate the fire from the lower cargo bay area up into the cabin. Comments by survivors regarding the appearance within the cabin of thick black smoke very early in the accident sequence are consistent with observations in other airplane accidents involving gaseous oxygen and fire. The Safety Board believes that the propagation of the fire in the cabin of USA1493 was accelerated by the release of oxygen from the flightcrew oxygen system that was damaged in the initial collision sequence on the runway and that the accelerated fire significantly reduced the time available for emergency evacuation. The Safety Board recognizes that gaseous oxygen systems are not required to meet specific crashworthiness standards and that there were unique impact forces resulting from this runway collision. The technical data surrounding this accident and the historical data regarding gaseous oxygen fires do not appear to be sufficient to support the need for specific airplane structural or systems modifications. The Safety Board is aware of and encourages ongoing FAA research on the potential for gaseous oxygen involvement in aircraft fires. The Safety Board supports this effort and urges the FAA to continue the research with a view toward system modification. 2.6.3 Adequacy of FAA Regulations Relative to Fire Retardant Cabin Furnishings The need for fire retardant cabin furnishings on transport aircraft was first addressed by the FAA in 1947. By 1972, FAA regulations required carpets, seats, and interior panels, to undergo Bunsen burner flammability tests. Subsequently, the FAA conducted additional research and proposed upgrading these standards by adding toxicity, smoke, and improved 68 flammability criteria. By 1977, in the absence of full-scale burn tests to support the rule and proposed standards, the rule was withdrawn. As a result, the FAA formed the Committee on Special Aviation Fire and Explosion Reduction (SAFER), which conducted full-scale tests and research and made recommendations for fire safety improvements. The technical information developed as a result of these tests provided a standardized method of evaluating the suitability of cabin materials. On April 16, 1985, the FAA issued a Notice of Proposed Rulemaking (NPRM) entitled "Improved Flammability Standards for Materials Used in the Interiors of Transport Category Airplane Cabins," which became a regulation. in 1985.

[ANALYSIS] The Safety Board believes that these factors reduced injuries and saved lives. The Safety Board also found that the rapid availability of adequate numbers of ARFF-trained fire fighters, from both Fire Station 80 and off-airport structural fire companies, allowed ARFF personnel to implement an interior fire attack immediately. Sufficient personnel also allowed the extrication of the first officer, while protecting him from fire. 65 During the emergency evacuation, the R-1 exit, the left and right overwing exits, and the R-2 exit were used. Many of the passengers stated that the cabin filled with thick black smoke within seconds of the impact with the building. It is possible that some of the passengers, who perished in the aisle waiting to exit through the row 10 exits, could have made their way aft to the R-2 door. However, based on survivors’ reports of the rapid infusion of thick smoke, it is more probable that the aft portion of the cabin became obscured by smoke early, limiting the use of the R-2 exit. The delay in opening the right overwing exit prompted by the passenger who "froze" and the subsequent altercation involving two other passengers significantly hampered the evacuation to the extent that additional passengers who may have been able to escape did not. The outboard seatback adjacent to the overwing exit, which folded forward and blocked part of the opening, also slowed the evacuation of passengers. However, it was not possible to determine the cumulative effect of these events. A deceased flight attendant and 10 deceased passengers were found lined up in the aisle from 4 1/2 to 8 feet from the overwing exits. They most likely collapsed while waiting to climb out the overwing exit. They perished as a result of smoke and particulate inhalation, strongly suggesting that they were able to make their way, possibly guided by the floor path emergency lights, to the overwing area from as far away as the forward cabin. 2.6.1 Flight Attendant Training and Performance The investigation included a review of USAir’s emergency procedures training methods and the use of cabin mockups for training. During initial emergency evacuation training, student flight attendants are required to evacuate a cabin filled with simulated smoke. The Safety Board determined that the "hands on" training was realistic and replicated (as much as possible in training) what could be expected in an actual emergency. However, based on the circumstances of this evacuation, three potential training issues warrant discussion. The airplane was equipped with personal breathing equipment (PBE). However, flight attendants are trained in accordance with FAA standards to use the PBE for fighting in-flight fires rather than as a supplemental breathing source in emergency evacuations. The deceased flight attendant, who found the L-1 exit inoperable, made her way down the center aisle to reach the overwing exit to facilitate passenger evacuation and to try to escape herself.

[ANALYSIS] ANALYSIS General ... ccc ccc cc ccc cece cece teen tee ee ee ee ee eee e ee eeeece Air Traffic co.cc ccc ccc cece cece cece cence eenes eee e ee eeee sees Airplane Conspicuity ........ ccc ccc cece cece ccc cette ence eees Flightcrew Situational Awareness and Vigilance ............... Communications PhraseOlogy ...... ccc cece cere eee e ee ee ee ecaee Survival Factors ....... cece ccc eee ee cee cece eee eee ee ee eceee Flight Attendant Training and Performance ...............-42.- Source and Migration of the Cabin Fire ............ cece eee ees Adequacy of FAA Regulations Relative to Fire Retardant Cabin FurnishingS ....... ccc cece cece ccc c cece cee e ese sceeseeees FAA Overwing Exit Row Regulations ............ cece eee eee eeee Improved Access to Type III Exits ......... cece ee eee ween eee Efforts to Reduce Runway IncursiOns ......... ccs esc cccececees Pilot Self-Medication ...... cc ccc cece cee c cece cece eee eeeeses Analysis of FAA Post-Accident Toxicological Testing .......... Cockpit Voice Recorder Reliability ........... cc cece cee ween eee CONCLUSIONS FINGINGS ... eee ee ccc eee eee eee eee eee teen eens eee e eee eeees Probable Cause 1... cc cece cc ccc cee cence teen eee eee e ee eeaees RECOMMENDATIONS ........... cece cece rece ce cect eee eect eee eeeees iv APPENDIXES Appendix A--Investigation and Hearing ............. cee ee enone 81 Appendix B--Personnel Information ........... ccc cece e cece teens 82 Appendix C--Extract From Cockpit Voice Recorder Transcript ... 85 Appendix D--CVR/ATC Recorded Data Correlation ................ 98 Appendix E--ASDE Equipment Outages ............. cece eee ee eee 102 Appendix F--Summary of Medical History of Colin F.

[ANALYSIS] The technical information developed as a result of these tests provided a standardized method of evaluating the suitability of cabin materials. On April 16, 1985, the FAA issued a Notice of Proposed Rulemaking (NPRM) entitled "Improved Flammability Standards for Materials Used in the Interiors of Transport Category Airplane Cabins," which became a regulation. in 1985. The regulation established new fire test criteria for type certification, required that the cabin interiors of airplanes manufactured after 1985, and used in air carrier service, comply with these new criteria, and required that cabin interiors of all other airplanes type certified after January 1, 1958, and used in air carrier service, comply with these new criteria upon the first replacement of the cabin interior. The accident B-737 was manufactured before the effective date of the regulation and therefore any retrofit of fire retardant cabin furnishings was required only in the event of a "general retrofit" by the carrier. Piecemeal replacements of cabin furnishings, except for fire-blocked seat covers, are not required to meet the new flammability standards. The FAA’s rationale for this policy was the adverse economic effect on the airline industry. Thus, it is reasonable to expect that if an air carrier applied this regulation, as written, an airplane in service for 20 or more years might never be subjected to a "general retrofit," which requires an upgrade to the fire retardant materials. In this accident, all of the cabin furnishings burned except for the carpeting and seats. The overhead bins melted and ignited and then fell on the passengers and the cabin floor. If cabin furnishings of the type specified for newly manufactured aircraft had been installed in the accident airplane, fire and toxic smoke might not have spread so quickly through the cabin. The Safety Board believes that after a specified date air carriers should be required to use fire retardant materials in all transport category airplane interiors that meet the provisions of 14 CFR 25.853. 2.6.4 FAA Exit Row Regulations On April 5, 1990, the FAA enacted the final rule for "exit row seating," which required all Part 121 and 135 operators to screen and brief passengers who are assigned seats in exit rows. The rule became effective on October 5, 1990. The Notice of Proposed Rulemaking, which was published on March 13, 1989, and the final rule provided only general guidance on how operators could comply with the rule by stating, "Airlines must take steps to inform passengers sitting in exit rows about what may be required of them in an emergency evacuation." Although this general guidance did not specify how operators were to comply with the rule, operators were required to have FAA-approved programs for procedures to screen and brief passengers. At the time of the accident, and almost 4 months after the final rule became effective, the FAA had not completed its review, approval or rejection of any of the programs submitted by USAir and 12 other operators.

[ANALYSIS] The difference between the estimates was likely due to the subjectivity in measuring the striation densities and interpreting the results. 77 At 1432:00, the CVR recorded a sound similar to the EICAS engine fire warning, and the FDR recorded the ENG FIRE R and master WARNING annunciations at the same time. These warnings occurred 15 seconds after the captain rejected the takeoff and 10 seconds before the airplane came to a stop. 78 At 1432:04, the CVR recorded the captain stating “ok, let’s run the uh checklist”; about 11 seconds later, the first officer stated, “checklist for engine fire.” Commanding the engine fire checklist was consistent with American Airlines’ special purpose operational training for rejected takeoffs and evacuations (see section 1.6.3.1), which both flight crewmembers had received. 79 The engine fire checklist applied to in-flight operations as well as ground operations. This issue is discussed further in section 2.2.2. NTSB Aircraft Accident Report 53 replied that he “pushed it twice.” The first officer then realized that he had pulled but not rotated the engine fire switch, at which time he accomplished that step. Because of the wind at the time (from 180° at 10 knots) in relation to the airplane’s location on runway 28R, the smoke was blowing away from the cockpit, and the flight crew could not readily see the amount of smoke coming from the right engine. Also, postaccident observations in an exemplar 767 airplane demonstrated that the first officer would not have been able to see the right engine and most of the right wing when looking out the cockpit right-side window from his seat. Thus, at that point, the flight crew was unaware of the severity of the fire.80 The flight crew was also unaware that the cabin was beginning to fill with smoke. About 4.5 seconds after the first officer discharged the fire extinguisher bottle in the right engine, the captain stated, “oh look at the smoke—check out the smoke.” The captain stated, during a postaccident interview, that he recognized that continuing the engine fire checklist would not be appropriate because the airplane was on the ground. He called for the evacuation checklist, which had nine steps, about 4 seconds after seeing the smoke. The captain stated that, while performing the evacuation checklist, he heard commotion outside the cockpit door and realized that the cabin was being evacuated. The fourth step of the evacuation checklist instructed the flight crew to cut off the fuel control switches to shut down both engines.

[CONCLUSIONS > FINDINGS] If the flight crew or the flight attendants had communicated after the airplane came to a stop, the flight crew could have become aware of the severity of the fire on the right side of the airplane and the need to expeditiously shut down the engines. 15. American Airlines did not adequately train flight attendants qualified on the Boeing 767 to effectively use the different interphone system models installed on the airplane during an emergency. 16. The Federal Aviation Administration’s inadequate actions to improve guidance and training on communication and coordination between flight and cabin crews during emergency situations, including evacuations, could lead to negative consequences for the traveling public if this safety issue continues to be unresolved. 17. The flight crewmembers and flight attendants did not coordinate in an optimal manner once the passengers were evacuated. 18. Evidence of passengers retrieving carry-on baggage during this and other recent emergency evacuations demonstrates that previous Federal Aviation Administration actions to mitigate this potential safety hazard have not been effective. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the failure of the high-pressure turbine (HPT) stage 2 disk, which severed the main engine fuel feed line and breached the right main wing fuel tank, releasing fuel that resulted in a fire on the right side of the airplane during the takeoff roll. The HPT stage 2 disk failed because of low-cycle fatigue cracks that initiated from an internal subsurface manufacturing anomaly that was most likely not detectable during production inspections and subsequent in-service inspections using the procedures in place. Contributing to the serious passenger injury was (1) the delay in shutting down the left engine and (2) a flight attendant’s deviation from company procedures, which resulted in passengers evacuating from the left overwing exit while the left engine was still operating. Contributing to the delay in shutting down the left engine was (1) the lack of a separate checklist procedure for Boeing 767 airplanes that specifically addressed engine fires on the ground and (2) the lack of communication between the flight and cabin crews after the airplane came to a stop. NTSB Aircraft Accident Report 77 4. Recommendations 4.1 New Recommendations As a result of this investigation, the National Transportation Safety Board makes the following new safety recommendations: To the Federal Aviation Administration: Establish and lead an industry group that evaluates current and enhanced inspection technologies regarding their appropriateness and effectiveness for applications using nickel alloys, and use the results of this evaluation to issue guidance pertaining to the inspection process for nickel alloy rotating engine components. (A-18-3) Require subsurface in-service inspection techniques, such as ultrasonic inspections, for critical high-energy, life-limited rotating parts for all engines. (A-18-4) Revise Advisory Circular (AC) 20-128A, “Design Considerations for Minimizing Hazards Caused by Uncontained Turbine Engine and Auxiliary Power Unit Rotor Failure,” based on an analysis of uncontained engine failure data since the time that the AC was issued, to minimize hazards to an airplane and its occupants if an uncontained engine failure were to occur.

[ANALYSIS] Remain Seated” to assure the flight attendants and passengers that the situation was under control. The guidance did not require captains to consult with flight attendants before making this announcement. Although the flight 383 captain called for the engine fire checklist 5.7 seconds before the airplane came to a stop, an announcement communicating the captain’s initial assessment that an immediate evacuation was not necessary would have provided awareness to the flight attendants of the flight crew’s intentions. During postaccident interviews, three of the flight attendants reported that, after the airplane came to a stop, they expected an announcement from the cockpit. It is understandable that the flight crew was focused on securing the right engine (given the engine fire warnings and the report of fire from ATC), and the NTSB recognizes that the flight crew’s performance of the engine fire and evacuation checklists was consistent with American Airlines’ training and procedures. Nevertheless, there would likely have been enough time, after the airplane came to a stop, for the captain to quickly instruct the flight attendants and passengers to remain seated. This statement could have then resulted in the flight attendants notifying the flight crew of the magnitude of the fire on the right side of the airplane and the need to shut down the engines. Such communication between the flight and cabin crews would have been effective crew resource management (CRM). Also, the QRH provided the following flight crew evacuation guidance: “the Captain must evaluate a specific situation, apply good judgment, and reach the best decision given the information available.” Communication from the flight attendants to the flight 84 According to the FAA’s website (www.faa.gov, accessed January 10, 2018), a SAFO “is an information tool that alerts, educates, and makes recommendations to the aviation community.” NTSB Aircraft Accident Report 58 crew regarding the conditions inside the cabin and outside the airplane would have helped the captain make an informed decision about the timing of the evacuation. The NTSB is also concerned that none of the flight attendants alerted the flight crewmembers about the evacuation. As previously stated, the flight attendants tried to delay the evacuation because they had not received evacuation instructions from the cockpit, but they initiated the evacuation, although the left engine was still running, due to the severity of the fire and the passengers’ panic. The American Airlines Flight Service Inflight Manual described the importance of having flight attendants update the captain if cabin conditions warrant an evacuation. Although the manual described the flight attendants’ authority to initiate an evacuation in a life-threatening situation without awaiting instructions from the flight crew, the manual also stated that flight attendants were to attempt to communicate with the flight crew before the evacuation if possible. Because one (or possibly two) of the seven flight attendants attempted to contact the flight crew via the interphone (but were unsuccessful, as further discussed in the next section), there was adequate time for the flight attendants to communicate with the flight crewmembers and alert them about the evacuation. During a postaccident interview, one flight attendant stated his opinion that it was not a priority to be on the phone if an engine fire occurred.

[ANALYSIS] What about the—.” At 1433:07, the first officer called “fuel control switches both cutoff.” NTSB Aircraft Accident Report 54 the fuel control switch on the affected side to shut down that engine. For an engine fire during flight, shutting down the affected engine only would be appropriate because the unaffected engine would be needed for continued flight; the items on the checklist after the first five memory items included actions related to configuring the airplane for landing. However, for an engine fire on the ground, the checklist did not include a step to shut down the unaffected engine or direct the flight crew to perform the evacuation checklist after the engine fire checklist memory items. Also, as stated previously, step 5 of the engine fire checklist consisted of rotating the engine fire switch to its stop, holding for 1 second so that a fire extinguisher bottle could discharge its contents (halon) to suppress the fire, and then waiting 30 seconds. If the engine fire warning light remained illuminated, the engine fire switch was to be rotated to its other stop and held for 1 second, which would discharge the second halon bottle. A Boeing fire specialist stated that the 30-second wait period in between discharging halon bottles was necessary because it would allow pilots time to deploy a secondary means to suppress an engine fire if it reignited due to hotspots within the engine core. The fire specialist also stated that the wait period would likely not be required on the ground; however, that information was not included in the checklist. American Airlines based the engine fire/engine severe damage or separation checklist on Boeing’s 767 checklist procedure for the same emergencies. Boeing’s engine fire checklist procedures for its other airplane models also appeared to be the same for an engine fire during flight or on the ground. However, other airplane manufacturers, including Airbus and Embraer, have a set of procedures for an engine fire on the ground and another set of procedures for an in-flight engine fire. American Airlines used both sets of procedures for its fleet of Airbus A319, 320, and 321 airplanes and Embraer E190 airplanes. The American Airlines A319/320/321 QRH “ENG (1 or 2) FIRE (On Ground)” checklist (dated July 5, 2016) had six immediate action items to be accomplished. Those items did not have any associated wait times. After the last item was accomplished, the checklist directed the flight crew to accomplish the evacuation checklist, if required. Similarly, the American Airlines E190 QRH (dated May 31, 2016) showed seven items under the “ENG (1 or 2) FIRE, Severe Damage or Separation” checklist. The first item referred the flight crew to the “Engine Fire on the Ground Checklist” (dated January 3, 2017), which had five items to be accomplished. Those items also did not have any associated wait times.

[ANALYSIS] Consistent with company procedures, flight attendant 1 picked up the interphone to notify the flight crew that he was going to initiate an evacuation, but he did not later remember if he performed the step to call the flight crew. Flight attendant 1 opened the 1L door at 1432:38 (28 seconds after the airplane came to a stop) and evacuated passengers via the 1L door slide. Flight attendant 5 opened the 1R door at 1432:42 (32 seconds after the airplane came to a stop) after flight attendant 1 initiated the evacuation, but she immediately blocked that exit because the engine fire would be too close to the 1R door slide. Flight attendants are trained to assess conditions outside of doors that they are operating by looking through a window in the doors. If all conditions are met (no smoke, fire, or airplane or ground debris), the doors can be opened, and the flight attendants can continue to assess the conditions outside of the airplane. It is possible that flight attendant 5 did not see the extent of the fire until the 1R door was fully opened because of the limited viewing area of the door window. Once the door was fully opened, she recognized that the fire would present a danger and took appropriate action to prevent passengers from using that exit. Flight attendants 2 and 3, who were assigned to the aft cabin doors (4L and 4R, respectively), saw fire on the right side of the airplane. Because she could hear the left engine running, flight attendant 2 initially blocked her assigned exit, which was appropriate. She tried to contact the flight crew on the interphone to report that the left engine was running but had difficulty using the interphone. She opened the 4L door at 1432:48 (38 seconds after the airplane came to a stop).83 Flight attendant 3 also blocked his exit because of smoke and fire outside the door, which was appropriate. He tried to use the interphone to make a PA announcement to calm passengers but also had difficulty using the interphone. (The difficulty that these flight attendants experienced with the interphone is discussed in section 2.3.2.1.) Although the actions of flight attendant 7 in assessing and opening the left overwing exit did not slow the evacuation at this exit, performing flight attendant 6’s assigned responsibilities was contrary to company procedures and training. Also, flight attendant 7 deviated from the evacuation procedures in the Flight Service Inflight Manual, which indicated that “prior to opening an exit, assess conditions outside to determine if exit and escape route [are] safe,” and began evacuating passengers from the left overwing exit. However, after opening the left overwing exit, flight attendant 7 should have recognized, from the sound of the engine, that the exit would not be viable for an evacuation. (The sound of the engine would have been the primary cue to flight attendant 7 that the engine was still operating.) Given the location of the left overwing exit relative to the left engine, flight attendant 7 should have blocked the exit until the engine was shut down.

[ANALYSIS] He called for the evacuation checklist, which had nine steps, about 4 seconds after seeing the smoke. The captain stated that, while performing the evacuation checklist, he heard commotion outside the cockpit door and realized that the cabin was being evacuated. The fourth step of the evacuation checklist instructed the flight crew to cut off the fuel control switches to shut down both engines. Although the right engine had already been shut down as part of the engine fire checklist, shutting down the left engine did not occur until the flight crew depressurized the airplane (the third evacuation checklist step), which the captain reported took a long time, even though the airplane had not yet been in flight.81 (The NTSB notes that, at this point, the exits would have been opened, so the cabin would have already been depressurized.) FDR data showed that the left engine was shut down about 59 seconds after the airplane came to a stop, and video evidence showed that the left engine spooled down about 10 seconds later. The NTSB concludes that the captain’s decision to perform the engine fire checklist was appropriate given his training, the information provided by ATC, and the fire warnings in the cockpit. However, the design of the engine fire checklist delayed initiating the evacuation checklist, shutting down the left engine, and commanding an evacuation from the cockpit. 2.2.2 Engine Fire Checklist Design American Airlines’ engine fire/engine severe damage or separation checklist for the Boeing 767 did not differentiate between an engine fire in flight or an engine fire while the airplane was on the ground. As stated previously, step 3 of the checklist instructed the flight crew to cut off 80 During a postaccident interview, the first officer stated that, after evacuating the airplane, he was “surprised” to see flames on the right side of the airplane because he expected to see a small fire. 81 Although only two of the nine items on the evacuation checklist are associated with depressurizing the airplane, the captain thought that the evacuation checklist was “slow and cumbersome” because of the “large portion” of time needed to depressurize the airplane. In an October 2017 e-mail, Boeing indicated that, before the airplane’s takeoff rotation, the cabin pressure would be less than 0.05 psig above the ambient pressure, and the amount of time needed to depressurize the airplane would be “very short.” The CVR transcript showed that, at 1432:57, the first officer called “cabin altitude mode selector manual?” to which the captain replied “OK.” At 1433:01, the first officer called “cabin altitude control hold in climb.” About 4 seconds later, the captain replied, “okay. What about the—.” At 1433:07, the first officer called “fuel control switches both cutoff.” NTSB Aircraft Accident Report 54 the fuel control switch on the affected side to shut down that engine.

[ANALYSIS] The guidance further stated that a minimum drip clearance distance of 10 inches from potential ignition sources of the engine nacelle, for static conditions, would be acceptable. Even though the size of the right wing dry bay and the drip clearance distance on the flight 383 accident airplane (a Boeing 767-300) were consistent with this guidance, the uncontained engine failure resulted in a subsequent fire. Postaccident examination of the airplane’s right wing revealed that an HPT stage 2 disk fragment (later identified as disk fragment A) penetrated the lower surface of the right wing beneath the dry bay, resulting in a hole in the dry bay.106 Disk fragment A also severed the main engine fuel feed line and caused a large breach and multiple small breaches between the dry bay and the fuel tank, which allowed fuel to flow into the dry bay and out of the hole of the wing. Video showed that a fire had already begun on the right side of the airplane as the airplane decelerated to a stop (as a result of the rejected takeoff). Although the quantity of fuel that spilled out of the wing as the airplane decelerated could not be determined, the released amount ignited, resulting in a fire that (1) damaged the right wing and its flight control surfaces, the right main landing gear, the right-side fuselage, and the right stabilizer and (2) caused the cabin to fill with smoke.107 Also, the NTSB investigated a September 2000 accident in which an uncontained engine failure on a Boeing 767-200 resulted in a subsequent fire.108 In that accident, the size of the dry bay in the main fuel tanks and the drip clearance distance were also consistent with the guidance in AC 20-128A. Thus, the accepted design precautions in AC 20-128A to reduce the risk of an uncontrolled fire for fuel tanks located in impact areas did not adequately minimize the hazards to the 767-200 and -300 airplanes from an uncontained engine failure. The NTSB concludes that future aircraft certification efforts would benefit from guidance on uncontained engine failure debris models and resulting design mitigations that is based on lessons learned from recent in-service events.

[ANALYSIS] Similarly, the American Airlines E190 QRH (dated May 31, 2016) showed seven items under the “ENG (1 or 2) FIRE, Severe Damage or Separation” checklist. The first item referred the flight crew to the “Engine Fire on the Ground Checklist” (dated January 3, 2017), which had five items to be accomplished. Those items also did not have any associated wait times. After accomplishing the last item on the checklist, the flight crew was directed to either “establish and communicate a plan” or “accomplish Evacuation.”82 American Airlines’ engine fire checklists that are specific to ground operations did not include the 30-second wait time between discharging a fire extinguishing bottle and determining if the second bottle would need to be discharged as well. That time could be critical in terms of containing a fire and/or commanding an evacuation. The NTSB concludes that engine fire checklists that specifically address ground operations would allow a flight crew to secure an engine 82 The third step on the American Airlines A319/320/321 QRH nine-step evacuation checklist instructed pilots to depressurize the airplane if the pressurization system was in the manual operating mode. The eighth step on the American Airlines E190 QRH 12-step evacuation checklist was to “push in” the “pressurization dump” switch light; Embraer airplanes have a guarded “dump” switch light, which, upon pressing, would allow the outflow valves to quickly open, depressurizing the airplane. NTSB Aircraft Accident Report 55 and command an evacuation, if required, in a timelier manner than engine fire checklists that do not differentiate between ground and in-flight operations. Therefore, the NTSB recommends that Boeing work with operators as required to develop and/or revise emergency checklist procedures for an engine fire on the ground to expeditiously address the fire hazard without unnecessarily delaying an evacuation. The NTSB also recommends that American Airlines, for all airplanes that it operates, review existing engine fire checklists and make changes as necessary to ensure that the procedures would expeditiously address engine fires occurring on the ground without unnecessarily delaying an evacuation. The NTSB notes that American Airlines’ evacuation guidance for flight crews was focused on assessing the need for an evacuation occurring after landing because of “a significant in-flight event” rather than assessing the need, after a rejected takeoff, for an evacuation due to an engine fire and announcing the captain’s intentions. American Airlines could mitigate this issue by implementing the recommended action. In addition, the NTSB recommends that the FAA, when approving the operating procedures of a 14 CFR Part 121 air carrier, require operators to develop and/or revise emergency checklist procedures for an engine fire on the ground to expeditiously address the fire hazard without unnecessarily delaying an evacuation. 2.3 Evacuation Issues 2.3.1 Evacuation Sequence All seven flight attendants reported that they remained in position at their assigned exits after the airplane came to a stop. Passengers moved quickly toward the exits and were urging the flight attendants to open them.

[ANALYSIS] However, after opening the left overwing exit, flight attendant 7 should have recognized, from the sound of the engine, that the exit would not be viable for an evacuation. (The sound of the engine would have been the primary cue to flight attendant 7 that the engine was still operating.) Given the location of the left overwing exit relative to the left engine, flight attendant 7 should have blocked the exit until the engine was shut down. Flight attendant 7 stated that he assumed that the left engine was still running but not at “full blast mode” because the airplane had come to a stop. He also stated that his main concern was getting passengers off the airplane because of the fire. However, the evacuation guidance 83 Flight attendant 2 decided eventually to allow passengers to evacuate from her assigned exit because the cabin was filling with smoke. However, she and flight attendant 3 had to hold passengers back until the slide stabilized; after deployment, the slide was blowing toward the back of the airplane because the left engine had not yet been shut down. NTSB Aircraft Accident Report 57 specifically indicated that “engine(s) still operating” was an unsafe condition. The one serious injury that resulted during the evacuation occurred after a passenger evacuated using the left overwing exit. Once on the ground, the passenger stood up to get away from the airplane but was knocked down by the jet blast coming from the left engine. The NTSB concludes that the flight attendants made a good decision to begin the evacuation given the fire on the right side of the airplane and the smoke in the cabin, but the left overwing exit should have been blocked while the left engine was still operating because of the increased risk of injury to passengers who evacuated from that exit. Therefore, the NTSB recommends that the FAA develop and issue guidance to all air carriers that conduct passenger-carrying operations under 14 CFR Part 121 regarding (1) discussing this accident during recurrent flight attendant training to emphasize the importance of effectively assessing door and overwing exits during an unusual or emergency situation and (2) providing techniques for identifying conditions that would preclude opening exits, including an operating engine. The NTSB notes that a safety alert for operators (SAFO) could be an effective means for conveying this information to affected Part 121 carriers.84 2.3.2 Flight and Cabin Crew Communication The American Airlines B767 Operations Manual, QRH guidance stated (in the General Information section) that, if an immediate evacuation is not required, the captain should make a PA announcement commanding “This is the Captain. Remain Seated. Remain Seated. Remain Seated” to assure the flight attendants and passengers that the situation was under control. The guidance did not require captains to consult with flight attendants before making this announcement. Although the flight 383 captain called for the engine fire checklist 5.7 seconds before the airplane came to a stop, an announcement communicating the captain’s initial assessment that an immediate evacuation was not necessary would have provided awareness to the flight attendants of the flight crew’s intentions.

[ANALYSIS] Details of the examination of each of these 48 NTSB Aircraft Accident Report 49 sources can be found in appendix C. However, none of these were proved conclusively to be a source of the combustible fluid. The NTSB concludes that a combustible fluid, such as oil, hydraulic fluid, or fuel, was available in the engine; however, fire damage precluded the determination of the specific source of the combustible fluid. 2.4 Flight Crew Performance As noted, during the flight, the pilots encountered an uncommanded opening of the ATSV followed by indications of an engine fire. During the brief flight, the pilots also encountered several other abnormal events, including electrical and hydraulic system anomalies and the nose landing gear’s failure to extend. The investigation revealed that the flight crew did not perform several of the appropriate checklists and interrupted an emergency fire-related checklist. This section discusses in detail the pilots’ responses to these anomalies and the possible reasons for them. According to CVR evidence, about 74 seconds after the takeoff roll began and while the airplane was climbing through about 1,500 feet mean sea level, the first officer detected the illumination of the ATSV-Open light for the left engine. About 1 minute later (1313:55), the Left Engine Fire aural warning sounded, and the Left Engine Fire warning light illuminated. A review of the CVR transcript revealed no evidence that the flight crew performed any of the L or R Start Valve Open checklist items during the 53-second period from the detection of the ATSV-Open light to the onset of the Left Engine Fire warning.59 This section details the pilots’ actions in response to both abnormal and emergency alerts and some of the possible reasons for and the effects of these actions on the events that occurred both in flight and after landing. Section 2.5 considers additional human factor issues that were identified during the investigation. 2.4.1 Detection of and Response to the Left Engine ATSV-Open Light The proper operation of the ATSV-Open light system could not be completely verified during postaccident inspections because the wiring harness within the nacelle sustained too much fire damage to confirm electrical signal continuity. The time that the ATSV-Open light illuminated could not be determined through aircraft inspection or CVR or FDR information. During the Before Takeoff checklist, the pilots checked the status of the annunciator panel, and they made no comments about the illumination of the ATSV-Open light at that time. 59 In postlanding statements recorded by the CVR and during postaccident interviews, the pilots indicated that they thought that only a few seconds elapsed between the illumination of the ATSV-Open light and the Engine Fire warning. NTSB Aircraft Accident Report 50 No direct evidence exists to determine exactly when the light illuminated during the 74-second period between the application of takeoff power and the first officer’s detection of the ATSV-Open light at 1313:02.

[ANALYSIS] About 1317:26, the CVR recorded the first officer stating, “I can’t even shut it off,” referring to the difficulty he was experiencing moving the fuel lever to the OFF position, which was the third item on the checklist. About 1317:36, the first officer read the next item on the checklist, which directed the nonflying pilot to pull the engine fire handle on the captain’s command. The CVR did not record the captain confirming that the first officer had pulled the fire handle; however, about 1317:52, the first officer advised the captain that both fire agents had been discharged, which is the next item on the checklist and could only have been accomplished if the fire handle had been pulled. Even though the checklist had not been completed, the first officer then announced that he was going to set the landing speed settings. CVR evidence indicated that the first officer was subsequently distracted as he tried several times to shut the cockpit door, which kept opening as a result of electrical problems, and responded to several ATC transmissions. The NTSB is concerned about how the pilots managed and prioritized tasks during this phase of the flight and is especially concerned about the flight crew’s interruption of the emergency Engine Fire/Damage/Separation checklist. During the 2 minutes that elapsed from the time that the first officer initially started the checklist to the time he resumed it, the first officer engaged in radio communications for about 63 seconds, and the captain engaged in the flight attendant briefing for about 31 seconds. About 80 percent of the pilots’ delay in performing the third item on the checklist, which would have shut down the fuel supply to the engine and alleviated the fire, was caused by their performance of tasks unrelated to shutting off fuel to the engine and stabilizing the fire. To minimize the severity of the fire, the pilots should have continued conducting, without interruption, the Engine Fire/Damage/Separation checklist up to, at a minimum, completing the following critical items: shutting off the fuel to the affected engine, pulling the associated engine fire handle, and discharging the fire agent. By delaying the performance of these critical items, the pilots exposed the passengers, cabin crew, and airplane to unnecessary risks. American Airlines stated that its pilots were trained in the simulator to complete the checklist to at least the point where fire agent was discharged before deviating from the checklist. Pilots are expected to know the possible consequences of interrupting or deviating from an emergency or abnormal checklist, especially if a fire is involved. Further, the CVR did 54 NTSB Aircraft Accident Report 55 not record the pilots completing the Engine/Fire/Damage/Separation checklist. Specifically, they did not start the APU, turn the fuel crossfeed on, or set the hydraulic system. About 1319:59, the flight crew attempted to extend the landing gear; however, the gear indicator lights did not provide valid information about gear status.

[ANALYSIS] Dismukes, “Concurrent Task Management and Prospective Memory: Pilot Error as a Model for the Vulnerability of Expert,” Proceedings of the Human Factors and Ergonomics Society 50th Annual Meeting (2006) pp. 901-913. NTSB Aircraft Accident Report 53 After the captain briefed the flight attendants, the pilots made remarks about the valve indications, and the first officer continued to handle communications with ATC. CVR evidence clearly shows that handling the radio communications affected the first officer’s ability to complete the Engine Fire/Damage/Separation checklist. The first officer could have taken steps to ameliorate the situation. For example, he could have told ATC to stand by so that he could complete the critical items on the checklist, or he could have handed off communications to the captain. However, as PIC, the captain should have recognized that handling the radio communications was making it difficult for the first officer to run the checklist and that he, the captain, could handle ATC communications without interfering with airplane control.64 More than 2 minutes after the captain interrupted the Engine Fire/Damage/Separation checklist, the CVR recorded the first officer resuming its performance as the captain took back control of the airplane. Before the third item could be acknowledged by the captain, the aural fire warning again activated about 1317:02. About 1317:16, as a result of the electrical anomalies caused by the fire, the captain detected a reverser-unlocked annunciator, the cockpit door opened, and the power transfer occurred, all of which distracted the pilots from the checklist. The increased time allocated to running the Engine Fire/Damage/Separation checklist as a result of interruptions also increased the pilots’ workload as the situation deteriorated. During the final interruption, electrical problems resulting from the engine fire caused some instrumentation loss, the illumination of multiple annunicators, and the aforementioned malfunctioning of the cockpit door, all of which increased the pilots’ workload. Coping with partial instrumentation increased the captain’s flying workload, and the multiple annunciators65 increased both pilots’ troubleshooting workload by creating confusion about the actual state of the airplane and forcing the pilots to constantly reassess the situation. Responding to the malfunctioning cockpit door also increased the pilots’ workload and distracted them during their efforts to configure the airplane for landing.66 As noted, the workload associated with ATC communications was also relatively high during this portion of the flight. After the accident, the captain described the workload during the event “[as] busy as he could handle at the time.” The first officer stated that it was an “extremely compressed period of time and they had a lot to do.” The pilots could have used various methods to manage the increasing workload but primarily used task shedding. For example, as noted previously, the pilots did not complete several checklists, most likely because they were focused on flying the airplane and communicating with ATC and because of the workload and stress associated with addressing the emergency situation.

[CONCLUSIONS > FINDINGS] The pilots might not have immediately detected the air turbine starter valve (ATSV)-Open light illumination because of its location, static appearance, and color, and, once they detected the light, the pilots did not immediately respond to it because an open ATSV was considered an abnormal situation that did not require immediate action and they were involved in air traffic control communications and airplane configuration changes. 12. Coupling the air turbine starter valve (ATSV)-Open light with the Master Caution system might increase pilots’ ability to detect the presence of an abnormal ATSV condition; however, unintended consequences, such as aborted takeoffs, may occur and more work needs to be done to determine whether the Federal Aviation Administration should mandate the modification of the ATSV-Open light in the MD-80 fleet. 13. The pilots failed to properly allocate tasks, including checklist execution and radio communications, and they did not effectively manage their workload; this adversely affected their ability to conduct essential cockpit tasks, such as completing appropriate checklists. 14. No preexisting indicators in the pilots’ training or performance histories were found that could explain their poor performance during the accident flight. 15. The pilots’ interruption of the emergency Engine Fire/Damage/Separation checklist at a critical point prolonged the fire and led to additional problems, including the loss of hydraulic pressure, which caused the nose landing gear to fail to extend. 16. Given the airplane’s altitude and the lack of time to prepare for a nose landing gear up landing, the captain’s decision to go around was a reasonable choice. 17. The captain’s decision not to conduct an emergency evacuation after the airplane landed was in accordance with company guidance and was appropriate because the fire was not severe and aircraft rescue and firefighting personnel were actively responding to the residual fire. 18. The incident commander’s decision to deplane the passengers after fuel spilled out of the engine area was prudent. 19. The first officer did not have a clear understanding of the relationship between the pneumatic crossfeed handle and the engine fire handle, most likely because of inadequate company 67 NTSB Aircraft Accident Report guidance and training on the issue; this resulted in the first officer inadvertently reintroducing fuel to the left engine, creating potential unnecessary risk of fire. 20. Improved pilot training methods for responding to multiple systems failures, competing task demands, and increased workload would help pilots develop the skills and decision-making abilities needed during both single and multiple abnormal and emergency situations. 21. The casual atmosphere in the cockpit before takeoff affected and set a precedent for the pilots’ responses to the situations in flight and after landing, eroded the margins of safety provided by the standard operating procedures and checklists, and increased the risk to passengers and crew. 22. Operational procedures requiring that an airplane be configured for an evacuation when it is stopped away from the gate after a significant event would help expedite an emergency evacuation if one became necessary. 23. During the emergency situation, the flight attendants did not relay potentially pertinent information to the captain in accordance with company guidance and training. 24.

[ANALYSIS] Several electrical ignition sources are contained within the nacelle, including electrical and ignitor box wiring and the ATS. However, these were inspected, and no evidence of arcing was found. The uncommanded opening of the ATSV and the subsequent ATS freewheeling and failure could be an ignition source. During high-power engine conditions, such as those that exist during takeoff and initial climb, the open ATSV would allow a high-velocity airstream of about 600° F to flow into the engine nacelle area. Once the turbine bearing has failed, the exhaust airstream could momentarily reach temperatures of about 2,000° F if the oil from the ATS gearbox is ignited. Additionally, small particles of incandescent metal, which would be more than 2,000° F, would be generated and ejected from the exhaust duct when the turbine contacts the shroud. Therefore, the NTSB concludes that the open ATSV and resulting failed ATS allowed a hotter than typical airstream and/or incandescent particles to flow into the engine nacelle area and likely provided the ignition source for the in-flight fire. 2.3.2.2 Combustible Fluid Source As noted, the freewheeling ATS allowed hot engine bleed air to flow into the engine area for more than 2 minutes. This hot air flow could serve as an ignition source in the presence of an exposed, combustible fluid; however, this event does not sufficiently explain why a fire started in the nacelle cavity. Components and equipment located in the area exposed to the hot exhaust of an open ATSV are required to be fire resistant and, therefore, capable of withstanding the temperatures present during an ATSV-open event and of preventing combustible fluid sources in the engine from being readily available to the ignition source. Accordingly, an open ATSV condition, even during high-power phases of flight, should not degrade and compromise fluid lines, fittings, or components containing combustible fluids located in this area of the engine nacelle. Historically, prolonged ATS freewheeling and contained ATS failures have been rare and, when they have occurred, they have typically caused only minor heat damage to parts in the area adjacent to the ATS exhaust duct. Investigators considered several possible sources for the combustible fluid including: a preexisting latent failure condition of a nacelle component; the degradation of an engine component; a small hole in the fuel or hydraulic lines; a hydraulic fluid leak; and a stainless steel fuel fitting exposed to hot exhaust from the ATS. Details of the examination of each of these 48 NTSB Aircraft Accident Report 49 sources can be found in appendix C. However, none of these were proved conclusively to be a source of the combustible fluid.

[ANALYSIS] NTSB Aircraft Accident Report 62 the airplane up to the point of commanding an evacuation. The NTSB commends American Airlines’ change of its operational procedures to address this safety issue. The NTSB concludes operational procedures requiring that an airplane be configured for an evacuation when it is stopped away from the gate after a significant event would help expedite an emergency evacuation if one became necessary. Therefore, the NTSB recommends that the FAA require that operators provide pilots with guidance requiring that pilots and flight attendants actively monitor exit availability and configure the airplane and cabin for an evacuation when the airplane is stopped away from the gate after a significant event to help expedite an emergency evacuation if one becomes necessary. 2.5.4 Flight and Cabin Crew Communication Issues CVR evidence and postaccident statements indicate that the flight attendants did not detect smoke or fumes in flight. However, during the flight, the two flight attendants seated in the aft cabin did discuss hearing some popping noises that they thought could be associated with the left engine, but they did not convey this information to the cockpit or the lead flight attendant. At the time that the noises were heard, the pilots were shutting down the left engine and using the fire-extinguishing agent, and, therefore, it is unlikely that this information would have changed the outcome because the information was consistent with the known situation. Regardless, the NTSB is concerned that the information was not conveyed to the cockpit, as required by proper CRM procedures and company guidance that all crewmembers provide pertinent information to the captain to help in decision-making. Further, after landing, the pilots did not actively seek information from the flight attendants because they believed that the flight attendants would pass any significant information to them.71 However, long after the fuel spill, during the debriefing on the ground, a flight attendant stated that she had smelled fuel earlier, but she did not pass this information to the cockpit when it happened, which was, again, inconsistent with the pilots’ expectations and with company guidance and proper CRM.

[ANALYSIS] Although the situation on the ground did not deteriorate, the NTSB is concerned that the pilots allowed their attention to be diverted away from monitoring ARFF’s progress during an unresolved situation. The presence of ARFF personnel might have played a role in the pilots’ belief that it was an appropriate time to discuss the flight and decompress after the stressful emergency situation. However, as PIC and per company procedures, the captain was responsible for the passenger and crew safety until they disembarked from the airplane. The pilots should have been remaining vigilant to situational changes and soliciting all relevant and available information until they were notified that the fire had been extinguished. Although the sterile cockpit rule does not technically apply to postlanding events, because of the uncertainty of the situation in this case, the NTSB considers the postlanding events before the fire was extinguished to be a critical phase of this flight. The NTSB has addressed issues related to nonpertinent conversation and cockpit discipline during several previous accident investigations. For example, in its report on the August 26, 2007, accident in which Comair flight 5191 crashed during takeoff,69 the NTSB concluded that the “flight crew’s noncompliance with standard operating procedures…most likely created an atmosphere in the cockpit that enabled the crew’s errors.” In addition, the NTSB determined that a contributing factor to the accident was the flight crew’s nonpertinent conversation during the taxi, which resulted in a loss of positional awareness. The report also discussed previous accidents in which cockpit discipline was reduced and attention to task-related activities was diverted. The report also cited industry data showing that pilots who intentionally deviated from standard operating procedures were three times more likely to 69 National Transportation Safety Board, Attempted Takeoff From Wrong Runway, Comair Flight 5191, Bombardier CL-600-2B19, N431CA, Lexington, Kentucky, August 27, 2006, Aircraft Accident Report NTSB/AAR-07/05 (Washington, DC: NTSB, 2007). NTSB Aircraft Accident Report 61 commit other types of errors, mismanage errors, and find themselves in undesired situations compared with pilots who did not intentionally deviate from procedures.70 In addition, the NTSB has previously issued safety recommendations regarding cockpit discipline and sterile cockpit adherence. On February 7, 2006, as a result of the Corporate Airlines flight 5966 accident, the NTSB issued Safety Recommendation A-06-7, which asked the FAA to direct the POIs of all 14 CFR Part 121 and 135 operators to reemphasize the importance of strict compliance with the sterile cockpit rule. On April 28, 2006, the FAA issued SAFO 06004 to emphasize the importance of sterile cockpit discipline.

[ANALYSIS] When the first officer asked the captain about task allocation, the captain told him to “run the checklist” and stated that he, the captain, would fly the airplane. The captain did not indicate who should handle radio communications, even though, as PIC, it was his responsibility to allocate tasks. About 3 seconds later, before starting the checklist, the first officer engaged in communications with ATC. During emergency conditions, task allocation should change to better manage workload and effectively resolve problems. The American Airlines AOM stated that, during an emergency, the captain should designate a flying pilot, who should not perform other duties that could detract from airplane control. However, the guidance did not explicitly address whether the flying pilot or the pilot monitoring should be responsible for radio communications during emergency situations. This ambiguity might have contributed to the misallocation of tasks during the accident flight, specifically, the first officer’s handling of radio communications while conducting the checklist. About 1314:50 (about 54 seconds after the fire warning sounded), the CVR recorded the first officer calling out, “autothrottle off,” which is the first item on the Engine Fire/Damage/Separation checklist. The first officer then called for the left engine throttle to be placed in the idle position, which is the second item on the checklist. Shortly after the captain moved the throttle, ATC contacted the flight crew, and the first officer interrupted performance of the checklist for about the next 38 seconds to respond to several ATC queries. Subsequently, without asking the first officer the status of the Engine Fire/Damage/Separation checklist, the captain transferred control of the airplane to the first officer because he wanted to brief the flight attendants, even though according to the checklist, this task was to be completed after all critical items, including shutting off the fuel supply to the engine and stabilizing any possible in-flight fire, had been accomplished. The sustained interruption of the checklist occasioned by the ATC communications provided an opportunity for the captain to proceed to a noncritical task and for both pilots to fail to recognize that the critical items on the checklist had not been completed. Research has shown that interruptions can distract pilots and impede their completion of tasks because it becomes more difficult for them to maintain an awareness of what steps remain to be performed.63 The NTSB notes that the circumstances did not warrant further interrupting the checklist to brief the flight attendants because the event occurred during the initial climb, which meant that the cabin was in its takeoff configuration with passengers and flight attendants in their seats and carry-on baggage and carts stowed. As a result, the time needed to prepare for an evacuation would have been minimal. 63 K. Dismukes, “Concurrent Task Management and Prospective Memory: Pilot Error as a Model for the Vulnerability of Expert,” Proceedings of the Human Factors and Ergonomics Society 50th Annual Meeting (2006) pp. 901-913. NTSB Aircraft Accident Report 53 After the captain briefed the flight attendants, the pilots made remarks about the valve indications, and the first officer continued to handle communications with ATC.

[ANALYSIS] During postaccident interviews, the captain stated that ARFF personnel gave him the sense that they had the situation under control. Further, both pilots stated that, because ARFF personnel were actively applying fire-extinguishing agent to the residual fire, they concluded that it was safer for the passengers to remain on the airplane. About 1337, ARFF personnel reported that the fire was extinguished. The NTSB concludes that the captain’s decision not to conduct an emergency evacuation after the airplane landed was in accordance with company guidance and was appropriate because the fire was not severe and ARFF personnel were actively responding to the residual fire. 56 NTSB Aircraft Accident Report 2.4.5 Retraction of the Fire Handle While waiting for the tug to tow the airplane to the terminal, the first officer opened the left pneumatic crossfeed valve to provide air to the cabin. Shortly thereafter, ARFF personnel informed the pilots that fuel had spilled out of the engine area onto the ground, and the first officer pulled the left engine shutoff valve. Subsequently, the IC recommended to the captain that the passengers be deplaned as a safety precaution, and the captain complied. All of the passengers deplaned without injuries. The NTSB concludes that the IC’s decision to deplane the passengers after fuel spilled out of the engine area was prudent. According to the Boeing MD-80 FCOM, opening the pneumatic crossfeed valve on the MD-82 airplane causes the associated engine fire handle to retract. In this case, opening the pneumatic crossfeed valve caused the left engine fire handle to retract and fuel to be reintroduced to the left engine area. Although the American Airlines AOM indicated that the fire handle was mechanically linked to the pneumatic crossfeed lever and pilots are told during training that pulling the fire handle will mechanically shut off the pneumatic crossfeed valve, American Airlines’ manuals and training did not indicate that opening the crossfeed valve causes the fire handle to retract, reversing the shutoff of fuel. The NTSB concludes that the first officer did not have a clear understanding of the relationship between the pneumatic crossfeed handle and the engine fire handle, most likely because of inadequate company guidance and training on the issue; this resulted in the first officer inadvertently reintroducing fuel to the left engine, creating potential unnecessary risk of fire. As a result of this accident, the FAA issued SAFO 08018, which explained the fire handle characteristics of DC-9, MD-80, and MD-90 series airplanes and the relationship of the engine fire handle and the pneumatic crossfeed valve. The SAFO recommended that operators review their training and operating manuals to ensure that the design and interrelationship of the systems affected by the fire handle are adequately explained. It further recommended that they add a caution to their checklists stating that opening the crossfeed handle will retract the fire handle and potentially reintroduce fuel to a fire.

[FINDINGS] Their actions were particularly effective considering the number of persons in affected apartments, the substantial aircraft wreckage, and apartment building and vehicles fires. Command and control was well organized and expedited the dispatching of injured persons through the early establishment of an ambulance control center near the impact site. 3.1 49 3. CONCLUSIONS Findings 1. Weather was not a factor in the accident. 2. Air traffic services were proper and did not contribute to the causes of the accident. The pilots were properly trained and qualified for the flight. 4. The flightcrew experienced an in-flight fire leading to a request for an emergency landing. The special mission wiring was not installed properly, leading to a lack of overload current protection. 6. The FAA Form 337s provided instructions for the correct installation, and the mission power modifications made by another operator on 3 of the 18 special mission Learjets were correct. 7. Neither the mechanic(s) who installed the wiring nor the mechanic(s) holding the inspection authorization, who approved the installation, noted the nonconformity with the FAA Form 337 in the installation on N521PA and 14 other Learjets modified by the operator. 8. The in-flight fire most likely originated with a short of the special mission power supply wires in an area unprotected by current limiters. 9. The fire resulted in false engine fire warning indications to the pilots that led them to a shutdown of the left engine. 10. The intense fire, which burned through the aft engine support beam in flight, can be explained by a compromised fuel line resulting from a battery explosion. 11. 12. 13, 50 The in-flight fire caused substantial damage to the airplane. . structure and systems in the aft fuselage and may have precluded a successful emergency landing. At the time of impact, the left engine was not producing power; and the right engine was producing at least flight-idle power. The City of Fresno police, fire fighting, and rescue responses, which were assisted by units from Fresno Air Terminal, were timely and effective. 51 3.2 - Probable Cause The National Transportation Safety Board determines that the probable causes of this accident were: 1) improperly installed electrical wiring for special mission operations that led to an in-flight fire that caused airplane systems and structural damage and subsequent airplane control difficulties; 2) improper maintenance and inspection procedures followed by the operator; and, 3) inadequate oversight and approval of the maintenance and inspection practice by the operator in the installation of the special mission systems. 52 4. RECOMMENDATIONS © As a- result of the investigation of this accident, the National Transportation Safety Board makes the following recommendations: --to the Federal Aviation Administration: Publish an FAA Special Airworthiness Information Bulletin that describes the circumstances of this accident, including the consequences of improper installation of the special mission wiring, where electrical power wires were unprotected by current limiters.

[ANALYSIS] The improper installation left portions of the wires unprotected. by current limiters. The two large-diameter DC power wires that were retrieved from the wreckage showed evidence of arcing and they were "welded" together in the area }} ) 43 unprotected by the current limiters. This evidence indicates shorting and unlimited current flow for an extended period of time while the airplane was airbome. The investigation revealed that the proper installation specified that the special mission wires were to be attached to fittings in the generator control panel, where they would have been protected from overload during shorts by current limiters. Instead, the wires had been routed directly from the battery charging bus to 150 amp limiters before being routed to the control relay. Segments of the wires were unprotected by current limiters for about 18 inches, and they were found fused together between the battery bus and the current limiters. As a result of the wiring arrangement, large amounts of current could travel to the ground. 3 Fire damage to the wire insulation precluded a determination of the reasons for the shorting of the wires; however, one of these wires may have been frayed, and a short to ground (probably an airplane frame) occurred. Nevertheless, the evidence is conclusive that while the airplane was airborne, a severe short occurred on the left side of the aft fuselage in the electronics bay. Without the protection of the current limiters, the short was not interrupted. The evidence suggests that the arcing probably ignited wiring insulation and other combustible materials on the left side of the electronics bay. This caused damage to adjacent components. However, melting of the aft engine beam, the loss of Teflon electrical insulation on the engine fuel computer harness, and holes burned in the steel shield on the cabin conditioning hose required an intense fire directed at these items that was farther to the left side of the airplane. A diffuse fire limited to the electrical insulation and other solid combustibles in this area is unlikely to cause the type of fire damage noted. A "torch-like" flame from a pressurized fuel leak would be consistent with the fire damage noted on these items. The fact that these heavily fire-damaged components were in the same general location in the electronics bay of the airplane is also consistent with a burning fuel leak from a pressurized system. It is possible that the arcing or dead short drew excessive current, causing a battery to explode; this is supported by the conditions of the batteries and tiedowns. Two of the battery tiedown bolts were nearly straight, indicating the likelihood that they were not restraining substantial battery mass at the time of impact. Battery explosion, specifically the left battery, could have compromised a fuel line. A fracture at the motive flow valve could have resulted in a high pressure (300 psi) spray that became ignited and impinged upon the aft engine support beam. The fact that the inlet fitting of the left motive flow valve was 44 gone and the threads inside the valve were undamaged is consistent with a leak at that location.

[ANALYSIS] The left engine fuel and hydraulic valves were found in the open or normal operating positions. Laboratory examination of the bulb filaments found that the left engine T-handle, which had been recovered in the stowed position, showed no evidence of stretching. These are not the positions to be expected for an engine that was shut down because of a fire indication. Based on the ATC transcript, the flightcrew did not talk about any attempts at engine restart(s). About 1 minute before impact, after the captain said, "We got an engine fire on the right side too, it 42 shows," the first officer asked the captain, "do we have any power?" The captain stated, "I'm not getting any response, man." It is possible that the first officer began the restart sequence on the secured left engine. Disassembly found that both power levers were in the full-forward position at impact, and an attempted restart on the left engine would have initially required that the left engine fire T-handle be pushed into the normal position. The result would have been to reopen the left engine fuel and hydraulic valves. The flightcrew may also have received a fire warning later in the flight as the fire continued to cause damage in the aft fuselage area. Their comment about "fire on the right side too, it shows" supports this possibility. There was no reasonable means for the flightcrew to observe fire in the aft fuselage from the cockpit. Consequently, their remarks about the location of the fire probably came from the cockpit engine T-handle fire warnings. Determination of the at-impact location of pointers and digits would have been valuable in determining what engine instrument information, if any, was available to the flightcrew immediately before impact, as well as whether or how electrical shorting and a fire in the aft fuselage affected cockpit instrumentation. However, laboratory examinations were not able to gain useful information from the instruments. Although the airplane appeared to be in a controlled, gradual, highspeed descent until just before it crashed, the tower recording of the pilots’ voices indicated that they were having difficulties controlling the airplane during the last portion of the flight, as well as in diagnosing mounting problems with the airplane. The airplane crossed the extended centerline of the runway, did not tum to final approach, and subsequently crashed in a nose-low left-wing-down attitude. It crashed nearly aligned with a wide residential avenue. Consequently, the Safety Board's analysis of this accident focused on the reasons for the in-flight fire and the reasons that the pilots were unable to perform a successful emergency landing. The analysis also examined the oversight of the maintenance and inspection of the airplane by the operator. 2.2 I The Fire The evidence found in the wreckage showed that the electrical power cables for the special mission equipment had not been installed in accordance with specifications. The improper installation left portions of the wires unprotected. by current limiters. The two large-diameter DC power wires that were retrieved from the wreckage showed evidence of arcing and they were "welded" together in the area }} ) 43 unprotected by the current limiters.

[ANALYSIS] Battery explosion, specifically the left battery, could have compromised a fuel line. A fracture at the motive flow valve could have resulted in a high pressure (300 psi) spray that became ignited and impinged upon the aft engine support beam. The fact that the inlet fitting of the left motive flow valve was 44 gone and the threads inside the valve were undamaged is consistent with a leak at that location. It is also possible that the alcohol tank pressurization tube was melted through by the fire early in the sequence. The tube was not found in the wreckage; there was extensive fire damage in the aft fuselage area, and many nearby components were consumed or found partially melted. However, any compromise of the tube, early in fire propagation in the tailcone area, would have resulted in a ready source of engine bleed air into this enclosed space. There was evidence of soot in the aft tail section, specifically around lightening holes between the tail cone and vertical stabilizer; however, there was no evidence of severe smoke exit points from the fuselage. Further, most witnesses saw either no smoke or only light gray smoke trailing from the airplane before impact. Increased pressure flow from bleed air, impinging into the tailcone area, would have created a hotter fire, resulting in less smoke than expected emitting from the fuselage. 2.3 Airplane Handling The reason the flightcrew was unable to successfully land the airplane could not be conclusively determined. The 2 minutes of tower-recorded ‘intracockpit communications between the pilots prior to impact failed to provide sufficient data to determine the controllability of the airplane. The installation of a CVR and/or an FDR would have facilitated the determination of the events that led to the unsuccessful attempt to land the airplane. The Safety Board has addressed improved requirements for the installation and upgrading of CVR and FDR requirements for many facets of the aviation system. Although the installation of CVRs and FDRs are certainly important for passenger flight operations, they are also important for unique flight operations of aircraft with special mission equipment that operate over populated areas. This issue will be examined in the future by the Safety Board, and additional safety recommendations regarding operations, such as the accident flight, will be developed as appropriate. . . It is possible that the in-flight fire caused sufficient damage to the airplane structures and systems to render the airplane only partially controllable. -Although examination of the wreckage did not reveal a definitive reason for the loss of control, there is evidence that severe fire damage in the aft fuselage area occurred while the airplane was airborne. 45 Regarding engine power, the first officer asked the captain about 1 minute before impact, "do we have any power?” and the captain replied, "I'm not getting any response, man." However, engine disassembly evidence showed that although there was no indication of power on the left engine at impact, and it was only "windmilling,"” there was evidence of at least flight idle power on the right engine.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] It could not be determined to what cxtent the pilot followed the recommended procedures for an engine failure or fire in-flight. ‘The indication that the left propeller was feathered at impact was not necessarily the result of the manipulation of cockpit controls; in case of loss of engine oil pressure the propeller would feather regardless of pilot action, if the loss occurred at sufficient pm. The focl and oil shutoff valves behind the firewall were cable controlled; duc to the stretching and separation of these cables during the airframe breakup, the postcrash position of these valves would be an unreliable indicator of pilot actions, The switch thay controls the boost pump in the left inboard fuel cell was not recovered, A fire-induced failure in the pressure live to the fuel selector would result in fuel spillage regardless of selector position, With regard to the propagation of the fire from the engine compartment into the wheel well, the available evidence suggests that the fire burned through the left engine cowling, and progressed around the firewall, and then burned through the wheel-well skin immediately below the right augmenter tube. In this manner the fire would have entered the wheel well, below the fuel selector valve installed on the rear of the firewall. The burning of one or more of the aluminum fuel lines in this area would lead to an intense, uncontrollable wheel-well fire. The direct exposure of the unprotected wingspar caps to the fire in the wheel well led quickly to hea weakening of these load-carrying structures. The most intense fire occurred at the rear spar location; this spar was probably the first to lose its structural integrity, The subsequent increase in the tensile loads on the heated lower forward spar cap resulted in an upward overload failure of the left wing. 2.2 Conclusions (a) Findings 1, The aircraft was certificated in accordance with existing regulations. Improper maintenance procedures were utilized during the installation of the cylinders on the left engine. The pilot was certificated and qualified for the flight. Fire in the left engine commenced with the separation of No, 5 cylinder :ssembly. Uncontrolled fire progressed from the engine nacelle into the wheel well. The left wingspar cap was weakened by heat and eventually caused the in-flight separation oF the lef wing, The aircraft structure of the Beech Model 65 lacks adequate fire protection. (b) Probable Cause The Nationa! Transportation Safety Board determines that the probable cause of this accident was an uncontrolled fire in the left wheel well which resulted in loss of structural integrity of the left wingspars.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] The melting of the aft bracket that attaches the first stage of the augmenter assenibly to the engine support, as well as the burning of the adjacent cowling, the fucl lines to the fuel-flow transmitter, and the generator harness, indicated that a severe fire was burning in the right, lower section of the engine nacelle. Although the pilot would have been aware of an eigine problem when the No. 5 cylinder ceased to operate or caused vibration, it is unlikely that he would have been aware of the initial engine ce Oe Ale NAb Cal MINE RIE A Sy mtg . berg . Sob abigd vate gra BANE A eke iteneS hcota yb ne lib Bethel SOAR ED IMT ONE Le Se Salih MRR END, Aiea td RUTOLIR WER MOK NORMAL NI RO INE ain TY AnH ee comet AAG pater FMM MUM iAH MBA OR we le eT a ee Se caaitier’ ta Sch ame ttle ae RS oh edhe itn dartacea Sogn Epedai dows 4 Aas et ce ene NAOT HAR TSS ORIG fire, Most of the smoke and flames would have been forced outside, under the wing, through the exhaust augmencter tubes. These indications of fire would not have been readily apparene frora che cockpit until the right side of the left engine cowl or the upper wingskin above the wheel well burned through. Based on witnesses’ report of unusual engine sounds, and the mutilated condition of the No. 5 piston, it appears that the pilot did not immediately secure the left engine when the No. 5 cylinder became inactive. Despite vibrations, some power would still have been available, which might have prompted the pilot to attempt a landing on the cleared, hard-surface runway at Fairfield. The left turn described by the witnesses scoms to support this speculation, although this also happened to be the direction of predominant asymmetrical thrust. It could not be determined to what cxtent the pilot followed the recommended procedures for an engine failure or fire in-flight. ‘The indication that the left propeller was feathered at impact was not necessarily the result of the manipulation of cockpit controls; in case of loss of engine oil pressure the propeller would feather regardless of pilot action, if the loss occurred at sufficient pm.

[(b) PROBABLE CAUSE] The left wingspar cap was weakened by heat and eventually caused the in-flight separation oF the lef wing, The aircraft structure of the Beech Model 65 lacks adequate fire protection. (b) Probable Cause The Nationa! Transportation Safety Board determines that the probable cause of this accident was an uncontrolled fire in the left wheel well which resulted in loss of structural integrity of the left wingspars. The whecl-well fire resulted from an uncontained fire in the engine - ese sete Ras Raed bein ek Se an ee AER TUNER 6 WARP ASC cake wah emaelneana. ibe ameett emnecneryetuaie pep necnerener ernie oae ee Me EA ED Lett WEEN AON ORES NEED Stites ters nee atietate te SOO RAM NTRS BRR ETON AI Rp ERM il compartment which, in turn, was initiated by separation of one of the engine cylinders due to the use of improper maintenance procedures. Excessi-e working hours may have contributed to the oversight by the maintenance personnel involved, ous deviation from the design premise that an aircraft, should be able to tolerate an cngine failure, atu even an engine fire, without immediately affecting the aircraft’s structural integrity. The Board's Safety Recommendation A-72-21 through 24, issued March 3, 1972, and addressed to the FAA Administrator was an expression of that concern. The Administrator responded to these safety recommendations on July 5, 1972, The response indicated full compliance with the Board’s recommendations. (See Appendix E.) 3. RECOMMENDATIONS This accident, as well as a similar one in Australia, a month earlier, demonstrates a seri- BY THE NATIONAL TRANSPORTATION SAFETY BOARD; fsf JOHN H. REED Saget ercoeyantenrncncien ARAN NTP ARE RAE od RRNA AMERY Fe rine, BEANE SATE HEEL P AAR AD TIRED = MMR AIEEE ARE E Chairman /s) FRANCIS H. McADAMS eee reg bate Member “ee I se neem AARNE ann Ries oneal nie NNER AE Is}} 'tSABEL A. BURGESS Meisber PAO AE OREN CHR MOIR Re HEL RT OURO KEEN ATA OIE /sf WILLIAM R. HALEY _. Member Louis M. Thayer, Member, was absent, not voting. August 30, 1972. se auasasdae: 55 henge . SRRIR RESTA RUSTE PAERY TICS EET TRS SS SeLmMeLkRY eke Mamma eT Ate On mH mT NUE NA HAs PRES NER MMT REE ORE I RO AIO AO BO OMRON SES ed ahis SSUES RRS eat AGRE had age Waitt Biers aan as aes APPUNDIX A INVESTIGATION AND HEARING

[ANALYSIS AND CONCLUSIONS > ANALYSIS] This could occur cnly if the nuts were not installed, or if they were installed but not tightened. The engine manufacturer reported that untorqued nuts have been observed to back off the studs in approximately 10 minutes engine operating time. The three recovered lower studs failed as a result of alteriating bending overloads produced by the cylinder pivoting about these studs with the nuts probably still in place. When these studs eventually failed, the cylinder was released and moved away from the crankcase, at the same time displacing the exhaust stacks away from the augmenter tube and probably against the cowl door. It is apparert that the engine manufacturer's cylinder teplacument procedures were not followed. Had the prescribed tightening sequence been followed, it is highly unlikely that any nuts would have been omitted or left untightened. It should be noted, however, that there are convincing reasons which attribute the improper maintenance procedures in this case to fatigue of the mechanics involved, rather than to their carelessness or incompetence. In order to have the aircraft ready for the Sunday morning departure, the two mechanics worked approximately 27 hours on the preceding Friday and Saturday. ‘They finished their task around midnight, Saturday. The error potential implied in such a situation is a direct reflection on the quality control of the airline’s management maintenance program, The reason for the left-engine fire was directly related to separation of the No. 5 cylinder. This separation allowed engine oil to be pumped and spilled into the engine compartment in the immediate vicinity of the right augmenter tube. In addition, the No, 5 intake manifold pipe probably separated from the central induction system, thereby releasing a combustible fuel-air mixture, Ignition could easily have been caused by the exhaust stacks of the Nos. 1 and 3 cylinders or by a broken ignition wire from the No, 5 cylinder. The melting of the aft bracket that attaches the first stage of the augmenter assenibly to the engine support, as well as the burning of the adjacent cowling, the fucl lines to the fuel-flow transmitter, and the generator harness, indicated that a severe fire was burning in the right, lower section of the engine nacelle.

[ANALYSIS] In addition, during the latter parts of the recovery, the pilot continued a:, use excessive eievator deflection that resuited in excursions between 0.6 G and 1.6 G. Although DAC recommends that the airplane not be retrimed following a high altitude, high spced loss of control, the pilot applied ANU trim during the climb. Several seconds later, the airplme continued to pitch ~p even though thz elevators had returned to neutral for about 5 seconds. ‘he Safety Board determined that the continuing pitch up morion when the elevator was returned to neutral was a direct result ofrhe pilcf retrimming th.: airplane. The continued increase in pitch and AOA contributed to the first stall break (sudden pitch downj. As the airplane pirched down, the pilot continued to increase the ANU elevator deflection. At 118 seconds, the pilot again applied neariy full ANU elevator deflection as the nose of the airplane was dropping during 28 a stall. Analysis of the data indicate that stall breaks also occurred two other times, I at 66 and IC4 seconds, although the elevator deflections were not as severe. The Safety Board notes that the pilot chose to ignore the stall warning system and had to override the .%pound control column force to maintain the airplane in a stalled condition for about 2 minutes and 45 seconds. Since the pilot stated that he was experiencing severe turbulence, it is reasonable to conclude that he did not recognize that the motion cues were the result of stall buffet that he induced. The Safety Board believes that the sequence of events demonstrates the need for further. training for pilots flying the Mn>-i 1 to address aircraft handling during turbulence encounters and recovey procedures. The pilot used excessive force in attempting to control the pitch, retrimmed the airplane during a high altitude recovery, ignored the stall warning throughout the recovery, thought he was experiencing severe turbulence, and inappropriately pulled back on the control column durhg the stall breaks. The investigation revealed that neither DAC nor China Airlines had addressed the issue of high altiktde upsets in their training or flight manuals before the incident involving flight CI-012. DAC ciid address the subject in an AOL issued on April 29, 1993, entitled "Unintentiona! Slat Dep!oyment During Cruise." Although the AOL wzs issued in response to an unintentional slat deployment during cmise, it aoes address some areas that are appropriate to turbulence enccxmers and recovery pracedures.

[ANALYSIS] The Safety Board believes that it would be difficult for a piIot to avoid stalling the airplane by applying small control inputs consistent with light conrrol forces while trying to recover from the roll upset. In addition, the Safety Board believes that pilots !.lust receive hands-on training to experience the light control forces consistent with a high altitude. high speed loss of control. Written and verbal warnings are not sufficient. in the accidenr involving China Eastern flighi 583, the Safety Baard determined that the pilot of the MD-11 used excessive control deflections and delayed control deflections as a result of responding to stall warnings. In that accident, two passengers received Fatal injuries and many passengers were seriously injured because the excessive and poorly timed elevator deflections resulted in several cycles of positive and negative G. The pilot of China Airlines flight (3-012 used much smaller deflections ddring the recovery, (except for the large elevator deflections during the srall break) thus preventing large negative 6 excursions which have the potential to produce serious or fatal injuries. The Safety Board notes rhat both the pilot of Cf-012 and the pilot of the China Eastern MD-i i accident believed that they were experiencing severe turbulence rather than recognizing that they were inducing buffet as a result of a stall. 30 Alrhough the events of the CX-012 incident are different than those of China &stem, the Safety Board believes that both cdses clearly ir?dicate that specific pilot training is needed to ensure that pilots can promptly recover from high altitude upsets without inducing severe acceleration loads or multiple stalls. That training should be comprehensive enou& so that pilots can differentiate between severe turbulence and stall buffet. The Safety Board concludes that the pilot of China Airlines flight CI-012 used more control thm desirable or necessary during the initial portion of the upset and throughout the recovery. The initial overcontrol was the result of the Iight control forces inherent in the MD-! 1 design. The pilot's response to the stall warning was also not appropriate. However, in contrast to other ME-I 1 high altitude upsets induced by turbulence encounters or inadvertent slat deployments, this pilot did not command excessive nose-down elevator deflections during the recovery. This prevented negative G-load excursions that typically result in serious injuries to occupants. 2.4 IMD-lliDC-10 Pitch Stability DAC provided data to the Safety Board showing thst, at the same weigfits and same percent CG, the stick force per G are very sirllilar for the MD-1 ! and DC-10. The data ais0 shows that the MD-11 can operate at CGs further aft than the DC-10, thds, at the aft CGs the control forces for the MD-1 I are lighter than the DC-IO.

[CONCLUSIONS > FINDINGS] The captain continued to exert back force on the control column and thus maintained a stall condition, resulting in further excursion into the buffet regime. 12. The stall buffet, which was encountered as the airplane approached and entered the stall, produced a dynamic load on the outboard elevators that resuIted in structura! overload and failure of portions of the outboard elevators. 13. The elevator skin separation probably resulted from overstress produced during the stail buffet. 14. Control of the airplane following the incident was not adversely affected by :he loss of portions of the outboard elevators. 15. Douelas - Aircraft Company has not demonstrated by flight tests MD-I I stall recovery from abrupt high altitude. high speed upsets, nor were they required to do so as part of the certification process. 16. The pi!ots did not receive trainhg to aid in recovering from high altitude, high sFeed upsets in the MD-11. 17. The pilots did not receive hands-on training that demonstrated the light control forces encountered when marually flying at high altitudes and at high speeds in the MD-11. 42 3.2 Probable Cause I The National Transportation Safety Board determines that the probabIe cause of this incident was the light control force characteristics of the MD-I 1 airplane in high altitude cruise flight. The upset was induced by a moderate Lateral gust and was exacerbated by excessive control deflections. Contributing to the incident was a lack of pilot training specific to the recovery from high altitude, high speed upsets in the MD- I 1 , 43 As a result of the investigation of this incident, the National Transporntion Safety dozrd makes the following recommendations: --IO the Federa! Aviation Administration: Require Douglas Aircraft Company to advise MD-I 1 operators of the potential for damage to the composite elevators if the airplane is operared beyond the limits of the design buffet boundary, and to infom these operators that pilots might perceive the stall buffet (and subsequent loss of control) encountered during high altitude, high speed upseas as severe turbulence. (Class 11, Priority Action) (A-94-37) Require inspection. using nondestructive ultrasound "A" scan inspection techniques, of composite elevators on MD-I 1 airplanes lh3t are known to have been operated outside the design buffet bomdary. !Class II.

[ANALYSIS] It usualiy causes variations in indicated airspeed. '*A nondirnensionai number that is dared to turbulence. Values less than one usually result in ignificant turbulence. Severe turbulence: turbulence that causes large. abrupt changes in altitude and/or attitude. 11 usudly causes large variations in indicated airspeed. Aircraft may be mm~entxiiy out of control. 12"Severe Turbclence and Maneuvering from Airline Right Records," by R.C. Wingrove 3nd R.E. Bxh. 3r. 11 27 induce changes in angle of attack that are independent of pitch, but elevator control inputs induce changes in angle of attack that are correlated to pitch.” Therefore, if an AOA time hisiory is correlated to the pitch attitude time history, then the zirplme is not affec?ed by turbulence and is considered controllable in the vertical axis. Time history plots of flight CI-012’s elevator, pitch attitude, and AOA reveal than the trends of the airplane pitch a&itude data closely follow the trends of the A0.4 and elevator deflection throughout most of the upset and recovery. Aileron and elevator control deflections commanded by h\e pilot resulted in excessive roll and pitch excursions, at least four aerodynamic stalls, and almost continuous stall v:aming activation for a period of about 2 minutes and 45 seconds. The initial deviation from cruise flight was the result of a lateral gust from the left side of the airplane. The gust pduced an ANR sideslip that resulted in the airplane naturaliy rolling right and yawing left. The autopilot disconnected, probably from excessive roll rate, and the pilot applied EWD wheel deflection to counteract the increasing right rGTi angle. As the RWD roll rate was arrested, the LWD wheel deflection was not reduced rapidly enough io prevent a roll angle of 25 degrees to the left. The pilot commanded excessive control column deflections concitmnt with the excessive wheel deflections. The control column deflections resulted in rqidly increasing AOA and pitch angles that produced a high speed acceleration of about 1-65 G for about 8 seconds. The airplane transitioned into a 7,009 feet-per-minute climb €or the nex: 30 seconds and slowed to the 1 G stall speed. In addition, during the latter parts of the recovery, the pilot continued a:, use excessive eievator deflection that resuited in excursions between 0.6 G and 1.6 G. Although DAC recommends that the airplane not be retrimed following a high altitude, high spced loss of control, the pilot applied ANU trim during the climb.

[ANALYSIS] The Safety Board is ware that the FA4 is conducting z Special Certification Review of the MD-I 1. The review was prompted by tbe upset incidents and accidents and subsequent safety recommendatims issued bv the Board. The FAA is examining the handling qualities ofthe MD- 1 1 related to exceeding the buffet boundary md rhe structure of !he eievator reh?ed t ~ ! the damage sustained during excursions beyond the buffet boundary. 40 3. CONCLUSIONS 3.1 Findings 1. n L. 3. 4. 5. 6. 7. 8. 9. i 0. The flightcrew was certificated and qualified for the flight. Tne airplane was certificated and maintained in accordance with applicable regulations. Tne airplane was aispatcned in accordance wifi company procedures and Taiwanese regulations. There were no 2ir traffic control factors in the cause of the incident. The airplane encountered moderate turbulence. Recorda! values of flight CI-012’s lateral acceleration, heading, and roll angle indicate that a lateral gust initiated the upset. The autopilot disengaged, probably because of excessive roll rate, during the lateral gust. FDR dara indicate that the airplme stalied at least four times before the recovery. The flightcrew’s reactions to the lateral gust exacerbated the siruation 2nd led to significant pitch a d airspeed deviations and the unset of the airpiane’s stall warning. Because of the aft center-of-gravity (CG) position at which the MD-I i airplane is designed to be flown in high-altitude cruise, the airpime operates at lower longitudinal stability margins. Since ?here 2re no compensatory changes in the airplane’s pitch control system, control forces are lighter than for most conventional transport airplanes while performing comparable maneuvers. Consequently, a pilot is more likely to overcontrol the MD-11 airplane during recovery from a turbulence upsct. This overcontrol can lead to excessive positive load factors that 41 can cause the aiqlane to enter stail buffet, and/or to excessive negative load factors that can lead to severe injuries to unrestrained passengers. 1 1. Upon approach to the stall, the MD- I 1's Longitudinal Stability Augmentation System introduces a nose-down pitching moment that requires a heavy control force to counter. The captain continued to exert back force on the control column and thus maintained a stall condition, resulting in further excursion into the buffet regime. 12.

[CONCLUSIONS] The airplane subsequently departed controlied flight and sustained damage to the left and right outboard elevator skin assemblies, portions of which separated from the airplane. The airplane was operating under the provisions of Title 14, Code of Federal Regulations, Part 129, as a scheduled passenger flight from Taipei, Taiwan, to Anchorage, Alaska. There were 246 passengers, 3 flightcrew members, 2 additional crewmembers, artd 14 cabincrew members on board, none of whom reported any injuries. The airplane con?inued on and landed uneventfully at Anchorage, Alaska. The National Transportation Safety Board determines that the probable cause of this incident was the light control force characteristics of the MD-11 airplane in high altitude cruise flight. The upset was induced by a moderate lateral gust and was exacerbated by excessive control deflections. Contributing to tfle incident was a lack of pilot training specific to the recovery from high altitude, high speed upsets in the MD- I I . Safety issues discussed in the report include 'he design and certification of the MD-11 airplane. Safety recommendations concerning these issues were made to the Federal Aviation Administration. Also, on November IO, 1993, the Safety Board issued several safety recommendsrtions concerning the MD- 1 I that were relevant to this incident. XATIONAL TRANSPORTATION SAFETY BOARD WASHIIVGTON, DX, 20594 AIRCRAFT INCIDENT REPORT IN-FLIGHT TURBULENCE ENCOUNTER AYD LOSS OF PORTIONS OF THE ELEVATORS CHINA AIRLIXES FLIGHT CI-012 McDONNELL DOUGLAS MD-11-P TAIWAN REGISTRATION B-150 ABOUT 20 MILES EAST OF JAPAN DECEMBER 7,1992 1. FACTUAL INFOR.MAT1ON 1.1 History of Flight On December 7, 1992, about 1036 Coordinated Universal Time,' a McDonnell Douglas MD-11, Taiwan registration B-150, China Airlines, flight CI-012, exountered moderate turbulence at flight level (E) 330. The airplane subsequently departed controlled flight and sustained damage to the left and right outboard elevator skin assemblies, portions of which separated from the airplane. The airplane was operating under the provisions of Title 14, Code of Federal Kegutatlons (CrK), Faan ILY, as a s c ~ e ~ ~ k ~ pa>sciigci ili, to Anchorage, Alaska. There were 246 passengers, 3 flightcrew memwrs, 2 additional crewmembers, and 14 cabincrew members on board, none Df whom reported any injuries.

[ANALYSIS] Alrhough there was nc dam.age to the airplane that prevented it from continued flight, the seriousness of the In-flight divergence from controlled flight, and the unusual mode of failure of th? elevators on a relatively newly designed airplane, gave cause for concern and prompted the Safety Board's investigation. It also provided the Safety Board with the opportunity to examine the current technology concerning composite structures and their use in state-of-the-art airplanes. The outboard sections 3f both the right ana left elevators exhibited similar separation signatures indicating that the failwes were produced by a symmetrical loading condition. The evidence indicated that the elevatars exhibited fracture, delarninatiorl, and disbonding of the upper right and lower left outboard skin panel assemblies with predominantly adhesive failure modes. Tke Safety Board considered sources of loads that could have causd ihe failures. Among the areas examined were weather, flightcrew actions, SinICturdl design, surface preparaticn, and Statistical analysis and design substantiation. 2s 2.2 Weather - Turbulence Winds at FL 330 were westerly at abut 88 hots. A maximum wind speed of about 155 knots occurred around FL 400. The tropopause was around 45,oM) feet. Based on data obtained from the Japan Meteorological Agency and McIDAS? it was determined that significant turbulence and up and down vefiical motions probably occured in the area of the incident at FL 330. Calculated values for vertical and horizontal windshears were conducive to turbulence of at least moderate inten~ity.~ Calculated Richardson numbers'' were also consistent with a turbulent atmosphere. Several PIREPS in the area indicated moderate to severe turbulence.I1 In addition, there is some evidence that significant convection was Clccarring in the area of the incident. FDR data show that the airplane was encountering moderate turbulence at the time of the upset, as defmed by fhe recorded G forces. Consequently, the Safety Board concludes that flight CI-012 encountered moderate turbulence that preceded the violent motions of the airplane. 2.3 Crew Actions The Safety Board analyzed the FDR data to determine how the turbulence and pilot reactions resulted in the loss of control sf the airplane. A study by the National Aeronautics and Space Ahninistmtion's (NASA's) Ames Research Centeri2 suggests that "analysis of the sijr)rt-kIIIl variations in elevator deflection and aircrafr pitch angle" reveal that "v&caE winds 'McIDAS: Man computer Interactive Data Access System. McIDAS is an interactive meteorological analysis and data management computer system that was developed md administered by the Science and Engineering Center at the University of Wisconsin, Madison, Wisconsin. 'Mod-rate turbulence: turbulence that causes changes in altitude andior attitude, but the aircraft remains in positive control at dl times. It usualiy causes variations in indicated airspeed. '*A nondirnensionai number that is dared to turbulence. Values less than one usually result in ignificant turbulence.

[ANALYSIS] The data ais0 shows that the MD-11 can operate at CGs further aft than the DC-10, thds, at the aft CGs the control forces for the MD-1 I are lighter than the DC-IO. Therefore, the Safety Board noted with interest ihnt Jata presented by NASA (see footnote in section 2.3), show that three of the four CL;XS with significant pilot-induced neguive maneuvering loads were DC-10 airplanes (the other was an A-310 airplane). In addition, the Safety Board is mare of 1 1 other cases of pilot-induced maneuver loads involving MD-11 airpla~ies. The Safety Board is cmcemed that the MD-i 1 has been involved in a disproportionate number of hign altitude upsets in which pilot-induced flight loads were excessive. During flight tests. FAA rest pilots subjectively determined that the control characteristics and forces are adequate for the line pilot to accomplish a specific maneuver. DAC test pilots acknowledge that the longitudinal control forces of an MD- 1 I are lighter than for other transport-category airplanes. In addition, the control forces are even lighter at high altitudes and hieh .- speeds. Further, DAC and FAA test pilots have staied that recovery from abrupt. high altitudz, high speed upsets is not examined during the certification 31 prrxess. Ai&oug:I, DAC has stalled the MD- 1 1 during controlled high althde high speed staifs, the skill levels required to recover from abrupt turbulence or pilot- induced stalls have not been fully explored. The Safety Board concludes that the MD-1 1's light control forces make recovery from hi$ altitude, high speed upsets difficult for the pilot. In ies report on the China Eastern accident, the Board stated that a review of the handling qualities of the MG-11 w s needed tc ensure that pilot responses to pitch attitude upsets do not result in hazardous piach oscillations, structural damage, or any other condition *at could lead to ansafe Kight. Safety Recommendation A-93-147 issued to the FAA on November IO, 1993, addresses this issue (see section 4). However, the Safety Board is also concerned that there are no specific certification requirements or flight test standards that address the issue of recovery from abrupt, high altitude, high speed upsets. The Board believes that the FAA should establish certification requirements for appropriate flight control handling chzracteristics, sgch zs stick force per G limits, and require flight demonstrations to ensure that pilots can safely recover from abrupt, high altitude, high speed upsets. 2.5 structural Design and Manufacturing Process Since the failure mode of the majority of bead/skin separation was found to be adhesive, the nature of the adhesive was analyzed.

[ANALYSIS] Higher density altitudes result in a reduction of aerodynamic (wing and propeller) and powerplant performance during takeoff and initial climb, and in a longer takeoff run and slower rate of climb. This reduced rate of climb might well prompt a person who was inexperienced with high density altitude takeoffs to raise the nose of the airplane in an attempt to increase the rate of climb, thereby further decreasing the airspeed. Therefore, the Safety Board concludes that the high density altitude and possibly the pilot in command’s limited experience with this type of takeoff contributed to the loss of airspeed that led to the stall. The Safety Board has been unable to determine which of the above factors, or a combination of factors, resulted in the reduction in the climb speed to below the stall speed. However, the Safety Board concludes that the pilot in command failed to ensure that the airplane maintained sufficient airspeed during the initial climb and subsequent downwind turn to ensure an adequate margin above the airplane’s stall speed, resulting in a stall and collision with the terrain. The Safety Board notes that the pilot in command had limited experience operating out of high density altitude airports, such as Cheyenne, and that this should have prompted him to be cautious. He had expressed concern about the predicted storm that was to move in from the west, and he had wanted to leave early enough to avoid the storm. Further, just prior to departure, the pilot knew the wind conditions reported by a pilot who had just departed. Accordingly, the Safety Board concludes that the pilot in command inappropriately decided to take off under conditions that were too challenging for the pilot trainee and, apparently, even for him to handle safely. Therefore, the Safety Board attempted to analyze the human performance factors that might have caused the pilot in command to depart under those conditions. These factors include the possible effects of fatigue, the emphasis placed on media events, and the desire to adhere to the programmed itinerary. 41 2.3 Human Performance Aspects 2.3.1 Fatigue The pilot in command’s sleeping schedule in the days before the accident flight may have led to fatigue. He received 6 ½, 6 ¾, and 5 ½ hours of sleep, respectively, in the 3 days prior to the start of the trip on April 10, compared to the 8 ½ to 9 hours of sleep that he typically received per night on weekends.41 On April 10, he awoke at 0330, earlier than his normal wake-up time. By midafternoon on April 10, during the fueling stop at Rock Springs, he told a witness of being tired. After arriving at Cheyenne, he telephoned his wife and said that he “was really tired.” There is evidence that people tend to underestimate their level of tiredness,42 so that when the pilot reported being “really tired” it probably reflected a high level of fatigue.

[CONCLUSIONS] The pilot in command, pilot trainee, and rear seat passenger (the pilot trainee’s father) were fatally injured. Instrument meteorological conditions existed at the time, and a visual flight rules flight plan had been filed. The flight, which was a continuation of a transcontinental flight “record” attempt by the youngest “pilot” to date (the pilot trainee), was operated under the provisions of 14 Code of Federal Regulations Part 91. The National Transportation Safety Board determines that the probable cause of this accident was the pilot in command’s improper decision to take off into deteriorating weather conditions (including turbulence, gusty winds, and an advancing thunderstorm and associated precipitation) when the airplane was overweight and when the density altitude was higher than he was accustomed to, resulting in a stall caused by failure to maintain airspeed. Contributing to the pilot in command’s decision to take off was a desire to adhere to an overly ambitious itinerary, in part, because of media commitments. The safety issues discussed in the report include fatigue, the effects of media attention and itinerary pressure, and aeronautical decision making. A recommendation concerning the circumstances of this accident and the importance of aeronautical decision making was made to the Aircraft Owners and Pilots Association, the Experimental Aircraft Association, and the National Association of Flight Instructors. Recommendations concerning aeronautical decision making and the hazards of fatigue and were made to the Federal Aviation Administration. NATIONAL TRANSPORTATION SAFETY BOARD WASHINGTON, D.C. 20594 AIRCRAFT ACCIDENT REPORT IN-FLIGHT LOSS OF CONTROL AND SUBSEQUENT COLLISION WITH TERRAIN CESSNA 177B, N35207 CHEYENNE, WYOMING APRIL 11, 1996 1. FACTUAL INFORMATION 1.1 History of Flight On April 11, 1996, about 0824 mountain daylight time (MDT),1 a privately owned Cessna 177B, registration N35207, collided with terrain after a loss of control following takeoff from runway 30 at the Cheyenne Airport, Cheyenne, Wyoming. The pilot in command, pilot trainee,2 and rear seat passenger (the pilot trainee’s father) were fatally injured. Instrument meteorological conditions existed at the time, and a VFR3 flight plan had been filed.

[CONCLUSIONS > FINDINGS] The accident sequence took place near the edge of a thunderstorm. 51 11. The pilot in command decided to turn right immediately after takeoff to avoid the nearby thunderstorm and heavy precipitation that would have been encountered on a straightout departure. 12. The airplane was 96 pounds over the maximum gross takeoff weight at takeoff, and 84 pounds over the maximum gross takeoff weight at the time of the impact. 13. Although horizontal in-flight visibility at the time of the stall was most likely substantially degraded due to precipitation, eliminating a visible horizon, the pilot in command could have maintained visual ground reference by looking out the side window. However, this could have been disorienting to the pilot. 14. The airplane experienced strong crosswinds, moderate turbulence and gusty winds during its takeoff and attempted climb, and the pilot in command was aware of these adverse wind conditions prior to executing the takeoff. 15. The right turn into a tailwind may have caused the pilot in command to misjudge the margin of safety above the airplane’s stall speed. In addition, the pilot may have increased the airplane’s pitch angle to compensate for the perceived decreased climb rate, especially if the pilot misperceived the apparent ground speed for airspeed, or if the pilot became disoriented. 16. The high density altitude and possibly the pilot in command’s limited experience with this type of takeoff contributed to the loss of airspeed that led to the stall. 17. The pilot in command failed to ensure that the airplane maintained sufficient airspeed during the initial climb and subsequent downwind turn to ensure an adequate margin above the airplane’s stall speed, resulting in a stall and collision with the terrain. 52 18. The pilot in command inappropriately decided to take off under conditions that were too challenging for the pilot trainee and, apparently, even for him to handle safely. 19. The pilot in command suffered from fatigue on the day before the accident. 20. Information on fatigue and its effects, and methods to counteract it, might have assisted the pilot in command to recognize his own fatigue on the first day of the flight, and possibly enhanced the safety of the trip. 21. The airplane occupants’ participation in media events the night before and the morning of the accident flight resulted in a later-than-planned takeoff from Cheyenne under deteriorating weather conditions. 22. The presence of media at the Cheyenne Airport and media interviews scheduled for the next two overnight stops probably also added pressure to attempt the takeoff and maintain the itinerary. 23.

[ANALYSIS] Casper Automated Flight Service Station (AFSS) provided a weather briefing to the pilot in command. A review of the briefing transcript revealed that the briefer adhered to agency guidelines and provided a comprehensive weather briefing to the pilot in command. As a result, the Safety Board concludes that the pilot in command was provided with a satisfactory weather briefing prior to departing Cheyenne. Although the pilot in command stated that he was unable to access the most recent ATIS broadcast on his radio, this was not a factor in the accident because he received all significant weather information from the tower controller and the AFSS weather briefer. 35 2.2 Analysis of the Accident Sequence 2.2.1 Operation of the Airplane’s Controls Autopsy findings of multiple wrist, ankle and feet fractures to the pilot in command indicate that he was handling the flight controls at the time of impact, and the absence of such injuries to the pilot trainee suggests that she was not handling the flight controls at the time of impact. It is unknown whether the pilot in command assumed control of the airplane just prior to the impact, or whether he was flying during the entire takeoff and accident sequence. The portrayal of the trip as an attempt to break the transcontinental trip record by the youngest “pilot” suggests that the pilot trainee would be doing the flying herself. Consistent with this portrayal, she was seated in the left seat in front of the flight and navigation instruments, which is the seat generally occupied by the person who is operating the controls. However, it is known that the pilot trainee had not, in fact, done all of the flying herself on the first day of the flight.39 Further, strong crosswinds, moderate turbulence, and gusty winds existed at the time of the takeoff. As indicated by the control difficulties experienced by the Cessna 414 pilot, who took off just before the accident flight, the weather conditions at the time the airplane took off were challenging even for experienced pilots, and it is unlikely that the pilot trainee could have handled them without some assistance. Therefore, the Safety Board concludes that the pilot in command was at least assisting the pilot trainee (if he was not the sole manipulator of the controls) during the takeoff and climb-out sequence, and, at the time of impact, the pilot in command was the sole manipulator of the airplane’s controls. 2.2.2 The Accident Scenario The statements provided by witnesses indicated that the airplane’s climb rate and speed were slow and that after the airplane transitioned to an easterly heading, it rapidly rolled off on a wing and descended steeply to the ground in a near vertical flightpath, consistent with a stall. The Safety Board 39The pilot trainee’s father told her mother at the end of the first day that the pilot trainee had slept for part of the trip, and that the pilot in command had assisted her in one of the landings because of strong winds. 36 analyzed the factors that may have increased the airplane’s stall speed and reduced its climb speed to the point at which it stalled.

[ANALYSIS] It is known that the precipitation at the time of the accident was at least partly, if not totally, frozen. Therefore, to the extent that the precipitation at the time of the accident sequence was frozen, the precipitation rate, and therefore the loss of lift, would be less. 37 witness statements indicate that the turn to the east was gradual; this is consistent with a bank angle of about 20 degrees. With the flaps at 10 degrees, a 20-degree bank angle would increase the stall speed about 3 mph, from about 59 mph for steady level flight to about 62 mph. Using the total weights of the clothed occupants, a full load of fuel, personal baggage and miscellaneous items (food and fluids) plus the airplane’s basic empty weight, the Safety Board concludes that the airplane was 96 pounds over the maximum gross takeoff weight at takeoff, and 84 pounds over the maximum gross takeoff weight at the time of the impact. The additional weight would have increased the airplane’s stall speed about 2 percent. The combined effect of the rain, the excess weight and the bank angle of the airplane could have increased the airplane’s stall speed to about 64 mph. The airplane’s best rate of climb speed at sea level is 87 mph (with a climb rate of 685 feet per minute). However, the high density altitude at the airport and the overweight condition of the airplane would have affected the airplane’s climb performance. The Cheyenne Airport has a field elevation of 6,156 feet msl. Taking into account the temperature, the density altitude at the time of the takeoff and accident was 6,670 feet msl. According to airplane performance data from Cessna, the high density altitude and the airplane’s overweight condition would have decreased the airplane’s best rate-of-climb speed to 81 mph; with a climb rate of 387 feet per minute. Thus, the combination of the effects of the rain, the overweight condition, and the gradual bank angle of the airplane would have increased the airplane’s stall speed from about 59 mph to about 64 mph, and, along with the high density altitude, decreased its best rate-of-climb speed from 84 mph to about 81 mph. However, the airplane should have been able to climb and turn safely. Thus, the Safety Board analyzed the possible reasons why this did not occur. These include a possible reduction in engine power from carburetor icing or an over-rich fuel/air mixture, and the effects of fluctuating winds, poor visibility and the lack of sufficient experience in takeoffs from high density altitudes on the pilot in command’s ability to operate the airplane. As noted in section 1.6.1, it is necessary to lean the fuel mixture at higher altitudes to allow maximum engine performance; an over-rich mixture can 38 result in an appreciable loss of power and reduced climb performance capability. The mixture control knob was found in the full rich (forward) position at the accident site, thus suggesting that it was in that position prior to impact, although it is possible that the knob was out and impact forces moved the knob forward without bending the rod.

[CONCLUSIONS > FINDINGS] The airplane occupants’ participation in media events the night before and the morning of the accident flight resulted in a later-than-planned takeoff from Cheyenne under deteriorating weather conditions. 22. The presence of media at the Cheyenne Airport and media interviews scheduled for the next two overnight stops probably also added pressure to attempt the takeoff and maintain the itinerary. 23. The itinerary was overly ambitious, and a desire to adhere to it may have contributed to the pilot in command’s decision to take off under the questionable conditions at Cheyenne. 53 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the pilot in command’s improper decision to take off into deteriorating weather conditions (including turbulence, gusty winds, and an advancing thunderstorm and associated precipitation) when the airplane was overweight and when the density altitude was higher than he was accustomed to, resulting in a stall caused by failure to maintain airspeed. Contributing to the pilot in command’s decision to take off was a desire to adhere to an overly ambitious itinerary, in part, because of media commitments. 54 4. RECOMMENDATIONS As a result of the investigation of this accident, the National Transportation Safety Board makes the following recommendations: --to the Aircraft Owners and Pilots Association, the Experimental Aircraft Association, and the National Association of Flight Instructors: Disseminate information about the circumstances of this accident and continue to emphasize to your members the importance of aeronautical decision making. (A-97-19) --to the Federal Aviation Administration: Expand the development and increase the dissemination of educational materials on the hazards of fatigue to the general aviation piloting community. (A-97-20) Incorporate the lessons learned from this accident into educational materials on aeronautical decision making. (A-97-21) BY THE NATIONAL TRANSPORTATION SAFETY BOARD James E. Hall Chairman Robert T. Francis II Vice Chairman John Hammerschmidt Member John J. Goglia Member George W. Black, Jr. Member March 11, 1997 55 5. APPENDIXES APPENDIX A INVESTIGATION AND HEARING 1. Investigation The National Transportation Safety Board’s Northwest Region was notified of the accident by the FAA’s Northwest Mountain Region Operations Center at 0758 Pacific daylight time on April 11, 1996. The Safety Board’s investigator in charge departed for the site on the first available commercial flight. Safety Board personnel from the South Central Field Office, Denver, Colorado, were the first investigators to reach the site, arriving shortly after noon. Investigation groups were formed for human performance and meteorology. Parties to the investigation were the Federal Aviation Administration, Cessna Airplane Company, and Textron Lycoming. 2. Public Hearing There was no public hearing held or formal depositions taken for this accident. 56 APPENDIX B TRANSCRIPT OF AFSS WEATHER BRIEFING Q Memorandum U.S.

[ANALYSIS] The pilot likely initiated a severe pull-up maneuver by pulling back on the control column; this action loaded the tail in the downward direction and the wing in the upward direction. The Safety Board concludes that the in-flight breakup occurred because the aerodynamic loading during the pilot’s 46 The Safety Board is not making any recommendations in this report regarding spatial disorientation because the FAA has already published many resources to acquaint pilots with the hazards of spatial disorientation. These resources include Advisory Circular (AC) 60-4A, “Pilot’s Spatial Disorientation”; Aeronautical Information Manual Paragraph 8-1-5, “Illusions in Flight”; Airplane Flying Handbook (FAA-H-8083-3) Chapter 9, “Flight by Reference to Instruments”; Instrument Flying Handbook (FAA-H-8083-15) Chapter 1, “Human Factors”; AC 00-6A, “Aviation Weather”; AC 61-23C, “Pilot’s Handbook of Aeronautical Knowledge”; and Medical Facts for Pilots (FAA-P-8740-41). In addition, the FAA has recently attempted to discover why the malfunctioning of attitude instruments in general aviation airplanes can result in loss of control in partial panel situations. See, for example, D.B. Beringer and J.D. Ball, “When gauges fail and clouds are tall, we miss the horizon most of all: General Aviation pilot responses to the loss of attitude information in IMC,” Proceedings of the 45th Annual Meeting of the Human Factors and Ergonomics Society (Santa Monica, CA: Human Factors and Ergonomics Society, 2001), Vol. 45, 21-25. 47 Research has demonstrated that visual distraction and head movements can often lead to spatial disorientation. See Benson, “Spatial Disorientation—Common Illusions,” 1999. 48 The closest official weather reporting facilities to the accident site—Denver International Airport, Centennial Airport, and Buckley Air Force Base—all reported that the lowest layer of clouds about the time of the accident was 600 feet. 49 See section 1.12 for more information. Analysis 31 Aircraft Accident Report pull-up maneuver was great enough to overload the horizontal stabilizer downward and the right outboard wing section was sufficiently loaded upward to sustain a permanent bend in the spar. As soon as the horizontal stabilizer fractured and departed the airplane, the right outboard wing loaded downward and separated from the airplane in a downward direction. 2.4 Loss of A.C. Electrical Power Four possibilities exist to explain the loss of a.c. power aboard the accident airplane. First, the selected inverter could have failed, and the pilot might not have switched to the other inverter. However, the pilot should have been familiar with this switch because it is always used to supply a.c. power after engine start and to terminate a.c. power before engine shutdown. Second, a dual inverter failure could have occurred. However, it is extremely unlikely that both inverters would have failed because of the inverters’ history of reliability aboard King Air 200 airplanes.

[ANALYSIS] Analysis 30 Aircraft Accident Report inner ear could have been stimulated by these head movements in addition to the motion of the airplane. Radar data, ground scars, and wreckage distribution indicated that the airplane impacted terrain at an angle that was less steep than the descent. The Safety Board concludes that the airplane’s angle at the time of impact indicated that the pilot attempted to arrest the descent in the final portion of the flight, possibly in response to obtaining visual references of the ground after emerging from the lowest layer of clouds.48 The pilot input and the airplane reaction required to arrest the descent rate with the available altitude would have placed a large aerodynamic loading on the airplane. The aerodynamic loading caused the airplane to break apart in flight at a low altitude (within several hundred feet of the ground) and crash into terrain. The ATC primary radar data returns about 1737:16, 1737:21, and 1737:26 (see figure 1) and the scattered debris found at the accident site were also consistent with an in-flight breakup at a low altitude. The cockpit and fuselage crown area showed severe impact damage, but the fuselage belly was not damaged. The airplane was found in an inverted position 125 feet from its initial impact point with the ground. Thus, the airplane initially impacted the ground nose first and then bounced 125 feet to an inverted position. 2.3 Breakup Sequence On the basis of the wreckage distribution,49 the Safety Board determined that the tail structure sections departed the airplane first and that the outboard section of the right wing departed the airplane after the tail sections. The pilot likely initiated a severe pull-up maneuver by pulling back on the control column; this action loaded the tail in the downward direction and the wing in the upward direction. The Safety Board concludes that the in-flight breakup occurred because the aerodynamic loading during the pilot’s 46 The Safety Board is not making any recommendations in this report regarding spatial disorientation because the FAA has already published many resources to acquaint pilots with the hazards of spatial disorientation.

[ANALYSIS] The Beechcraft Super King Air 200 Pilot’s Operating Handbook also indicates that, in the event of an electrical system failure, nonessential circuit breakers are not to be reset in flight, but essential circuit breakers are to be reset.44 However, there was insufficient evidence to determine whether either pilot attempted to troubleshoot the electrical problem by examining the circuit breaker panel, which was located on the right side of the cockpit at knee level. Despite the loss of a.c. electrical power, the pilot could have safely flown and landed the airplane from the left seat by referencing the available (non-a.c.-powered) flight instruments on the right side of the cockpit (the altimeter and the airspeed, attitude, and turn and slip indicators). Also, the pilot could have asked the second pilot to fly the airplane because the available flight instruments would be more easily viewed from the right seat. The Safety Board could not determine from the available evidence what actions the pilot took (or did not take) and the extent to which the pilot might have coordinated with the second pilot. Nevertheless, the Safety Board concludes that the pilot did not appropriately manage the workload associated with troubleshooting the loss of a.c. electrical power with the need to establish and maintain positive control of the airplane.45 Because mode C data were unavailable after about 1735:44, the Safety Board’s airplane performance study for this accident was used to estimate the airplane’s altitude and descent rate after this point. According to air traffic control (ATC) radar data, the airplane began to turn to the right about 1736:26. The calculated bank angle during this turn, estimated to be less than 15º initially, increased to more than 80º over the next 30 seconds. Also, airplane performance calculations showed that the airplane entered into a descent shortly after starting the turn to the right. The estimated descent rate increased constantly to more than 15,000 feet per minute as the airplane completed a 360º turn. The Safety Board concludes that the airplane’s estimated flightpath in the final 2 minutes of flight was consistent with a graveyard spiral resulting from pilot spatial disorientation.46 The pilot probably did not sense the right descending turn at first because the airplane’s bank was entered gradually. As the airplane’s bank angle and descent rate began increasing, the pilot’s spatial disorientation most likely persisted, and he was not able to successfully use the available instruments to regain positive control of the airplane. Any head movements that the pilot made while attempting to diagnose the electrical system malfunction (for example, looking down and to the right at the circuit breaker panel)47 might have exacerbated his spatial disorientation because the motion sensing organs of the 44 Because all of the circuit breakers were found open, most likely because of impact forces, the Safety Board could not determine the positions of the circuit breakers at the time of the accident. 45 In a postaccident interview, the air safety inspector who conducted the pilot’s March 1998 Part 135 check ride indicated that the pilot tended to “lock in on a problem and not fly the airplane.” This observation is consistent with the circumstances of this accident.

[CONCLUSIONS > FINDINGS] 3.1 Findings 1. The pilot was properly certificated and qualified under Federal regulations. No evidence indicates any preexisting medical or behavioral conditions that might have adversely affected the pilot’s performance during the accident flight. 2. The second pilot was properly certificated and qualified under Federal regulations. The second pilot was not a required flight crewmember because the airplane was certified for single-pilot operation under 14 Code of Federal Regulations (CFR) Part 91. Even if the flight had been operated under 14 CFR Part 135, the second pilot would still not have been a required crewmember; the airplane was certified for single-pilot operation under Part 135 in instrument flight rules conditions because a three-axis autopilot was installed and operating. Because the flight was conducted with two qualified pilots and an operational autopilot and thus exceeded Part 135 requirements, the circumstances of this accident would not have been any different if the pilot had operated the flight under Part 135 rather than Part 91. 3. The accident airplane was properly certified, equipped, and maintained in accordance with Federal regulations. The recovered components showed no evidence of any preexisting structural, engine, or system failures. 4. Icing was not a factor in this accident. 5. No evidence of a cabin pressurization problem was found. 6. This accident was not survivable for any of the airplane occupants because they were subjected to impact forces that exceeded the limits of human tolerance. 7. The physical evidence recovered from the wreckage site and the recorded radar data indicate that a complete loss of a.c. electrical power occurred aboard the airplane. 8. The pilot would have had salient cues to identify the a.c. electrical power failure. 9. The pilot did not appropriately manage the workload associated with troubleshooting the loss of a.c. electrical power with the need to establish and maintain positive control of the airplane. 10. The airplane’s estimated flightpath in the final 2 minutes of flight was consistent with a graveyard spiral resulting from pilot spatial disorientation. Conclusions 37 Aircraft Accident Report 11. The airplane’s angle at the time of impact indicated that the pilot attempted to arrest the descent in the final portion of the flight, possibly in response to obtaining visual references of the ground after emerging from the lowest layer of clouds. 12. The in-flight breakup occurred because the aerodynamic loading during the pilot’s pull-up maneuver was great enough to overload the horizontal stabilizer downward and the right outboard wing section was sufficiently loaded upward to sustain a permanent bend in the spar. 13. The a.c. electrical failure was a contributing factor to this accident but was not a causal factor because non-a.c.-powered instrumentation remained available for the duration of the flight for the pilot to use to safely fly and land the airplane. 14. Oklahoma State University did not provide any significant oversight for the accident flight. 15.

[ANALYSIS] The recovered components showed no evidence of any preexisting structural, engine, or system failures. (The in-flight electrical system failure is addressed later.) Snow was falling at Jefferson County Airport (BJC) in Colorado when the accident airplane departed for Stillwater Regional Airport (SWO) in Oklahoma, and instrument meteorological conditions (IMC) prevailed from the surface to altitudes above those of the accident airplane. However, icing was not a factor in this accident. No pilot reports surrounding the time of the accident indicated any in-flight icing over Colorado, and none of the pilot witnesses indicated any in-flight icing along or near the accident airplane’s route of flight. Also, radar data for the accident airplane indicated no degradation of airplane performance (airspeed or altitude) consistent with ice accretion. 39 This certification information is found in the Beechcraft Super King Air 200 Pilot’s Operating Handbook, Section II, “Limitations,” dated February 1979. Analysis 27 Aircraft Accident Report No evidence of a cabin pressurization problem was found. Also, the repeating cuts, rips, and holes on the bag included in the recovered oxygen mask and hose assembly were consistent with a bag that remained folded and unused. This accident was not survivable for any of the airplane occupants because they were subjected to impact forces that exceeded the limits of human tolerance. This analysis discusses the accident sequence, including the role of spatial disorientation. The analysis also describes the airplane breakup sequence and provides possible explanations for the loss of mode C altitude transponder information aboard the airplane. In addition, the analysis compares Oklahoma State University’s (OSU) revised team travel policy with the one that was in place at the time of the accident. 2.2 Accident Sequence Radar data indicated that the flight from BJC to SWO was routine until the airplane was at an altitude of 23,200 feet. At that point, the airplane stopped transmitting mode C altitude transponder information, which is generated from electrical output signals from the airplane’s a.c.-powered air data computer. The airplane continued to transmit mode A identifying transponder information, which is generated from d.c. electrical power, until the time of impact. Thus, a total power loss aboard the airplane did not occur, but a partial or complete loss of a.c. power had occurred. It was possible that the a.c. electrical malfunction involved a failure of the air data computer only. However, the pilot-side altimeter indicated that the altitude reading was 23,220 feet, which was consistent with the last reported altitude (23,200 feet) before the airplane stopped transmitting mode C information. Also, both radio magnetic indicator (RMI) compass cards showed a heading of 115º, which was consistent with the airplane’s heading when mode C information was lost. (The pilot was flying the airplane on a 110º heading and indicated that he was going to be making a turn of about 3º.) In addition, witness marks on the volt/frequency meter showed that it was at the 380-Hertz/100-volt position (the lowest reading) at the time of impact,40 indicating that no a.c. power was available.

[ANALYSIS] If the flightcrew did not react quickly and appropristely, it elso may have been impossible to recover from such a maneuver. According to the SCR report, the stall speed margins in many of the unmodified wing Model 24 aircraft have been found t) be inadequate, Maintenance of the Stall warning and pusher system is therefore critica: to the safety of flight. It is possible that this system in the accident aircraft may not have been properly adjusted since the aircraft had not been recently inspected in accord:ace with the manufacturer's recommended or FAA approved maintenance program. The Safety Bord could not conclusively rule out the possibility of flightcrew incapacitation as a factor in this accident because of the previous reported cabin pressurization problem. However, only about 1 minute 46 seconds elapsed between the time of the initial altitude excursion and the uncontrolled descent from FL 450. The aircraft descended 1 minute after this initial altitude excursion which is believed to have been caused by an encounter with clear air turbulence, Additionally, the evidence Suggests that at some point during the upset and descent, the flightcrew deployed the spoilers. Therefore, the Safety Board, believes it was unlikely that the uncontrolled descent was caused by flightcrew incapacitaticn., Since the Learjet hes characteristics which could lead to critical contro}} problems in the high altitude, high speed regime of flight, complex compensating features were Incorporated into the flight control system or required by Federal aviation regulations to provide for an appropriate leve) of safety. The integration of these compensating features vith the aircraft's primary flight control system requires strict adherence to sound maintenance practices to operate the aircraft safely. The minimum maintenance, accerded the aircraft while it was rapidly changing hands could have compromised this level of safety. Owners and operators must familiarize themselves sufficiently with newly acquired aircraft, Feders! reguiations, and maintenance programs to insure that aircraft are properly inaintaiied. This responsibility also extends to insuring compliance with all pertinent airworthiness directives and acquiring all pertinent service bulletins, flight manual, and service manual revisions, as appropriate. Because the AFM recovered from the wreckage did not include any of the current vevisions and the revisions were not located elsewhere in the wreckage, the Safety Board believes that tne flighterew probably did not have a current AFM aboard the aircraft. Although this ~23- Suggests that they may not have been aware of the relatively recent changes in the AFM restricting the use of snoilers, this fact could not be verified. The portion of the manual recovered from the wreckage was 4 copy of a Mode) 24 flight manual and the amendment concerning the warning about not deploying the spoilers above Vuo/Mmo could not be ound. 2.4. Training Complementary to proper maintenance practices in assuring flight safety of 6::7 alreraft are proper operational practices based on thorough pilot training and maintaining flyity proficiency. Thorough pilot training and a high level of flying proficiency are essential if the Learjet is to be operated safely.

[ANALYSIS] However, there were continuous pilot reported discrepancies concerning the autopilot which the Safety Board believes were not associated with the previous maintenance performed. The discrepancies concerned roll oscillation, “wandering” with the heading and altitude hold modes engaged, and the yaw damper’s possible contribution to these control difficulties. The discrepancies could have constituted a nuisance in flight and most likely resulted in the pilots avoiding the use of the autopilot. It was reportedly placarded inoperative by a pilot who delivered the aircraft to the current owner. A reported aileron misrigging could be attributed to the change in flight control surface balance after the aircraft had been painted. ‘i..ce a review of the maintenance records for June, July, and August 1981 did not disclose that any of the previously reported open discrepancies had been corrected, the Safety Board concludes that these problems probably continued to exist and that the aircraft had not been properly maintained since April 1981. The previously reported intermittent pressurization problem could have been a factor in the accident. Although the autopilot discrepancies could have caused thie flightcrew to avoid using it, there is no speed restriction on operation without an autopilot es there is with later model Learjets, which have a speed limit of 0.78 M,- However, use of the autopilot in turbulent air would assist in stabilizing the aircraft. Additionally, if the yaw damper had failet or malfunctioned, control of the aircraft could have been extremely difficult under turbulent conditions. ay . 2.3 Loss of Control The area in which the aircraft was transiting at PL 450 was characterized by a confluent zone of polar and subtropical jet streams. Based upon 2-minute average winds, the upper air sounding at Amarillo showed wind shear changes of greater than 6 knots per 1,000 feet at FL 450. The wind shears east of the aircraft's course over Dodge City were slightly less than 6 knots per 1,000 feet. Also, the aircraft was well within 6,000 feet of the tropopause, 4 transition zone between the troposphere and stratosphere, and a region where clear air turbulence is likely to be encountered. The weather pattern would have been conducive to moderate, and possibily severe clear air turbulence at FL 450. Considering these conditions and the accepted guidetines for turbulence forecasting, the Safety Board believes that a turbulence forecast should have been issued with the aviation area forecast. A turbulence SIGMET is not generally issued unless a pilot reports encountering turbulence, and there were no pilot reports of turbulence for 3 hours before the accident. However, it should be noted that there was no other traffic in the area at FL 450 for at least 30 minutes before the accident. The radar computer data showed that the aircraft was flying level at FL 451 and on course for 2 minutes before there was a disturbance In Its cruise altitude at 1456:21 and the aircraft suddenly climbed 100 feet.

[ANALYSIS] Additionally, a sharp, unexpected turbulence enzounter can easily cause an aircraft to exceed these margins. Although initial buffet margins can be exceeded, it does not necessarily mean that control difficulties will be immediately encountered. The degree to which the margins are exceeded will determine the aircraft's reaction. Tr - accident aircrafi's buffet margin was 0.14 M,, or about 41 KIAS in 1.5g flight. However, since the aircraft was operating in this relatively nerrow area of its flight envelope, a loss of control could haye occurred from a transient condition which might have placed the aircraft either below its low speed or above its high speed buffet boundary. This situation woild most likely have occurred if the flighterew had been inattentive (even momentarily) and did not take tir:ely and proper corrective action. Because the Safety Board was not ab!2 to determine ‘me magnitude of the potential gust factors involved, it is not possible to determine which buffet boundary would have been crossed in the turbulence encounter, Both boundaries were susceptible. However. a loss of control from either situation could result in a high speed uncontrollable descent if the pilot reacted inappropriately. Considering the phenomenon of the existirg weather, a gust upset of sufficient intensity could result in an overspeed and in control difficulty. Based on the FAA's SCR report, Mach tuck can vccur prior to M___ and alieron "buzz" can be encountered fust beyond this speed (0.82 M,). A production ror in the vopilot's pitot-static system or an error caused because the btatic sources were not flush with che fuselage, o.g., as & result of the recent painting, could be contributing factors leading te an overspeed. Such errors wealld have affected the proper operation of the stick puller and overspeed warning hori. If the flightzrew had been inattentive even mcmentarily and the aircraft had been allowed to accelerate beyond M » abnormal pitch forees, and a severe roll control problem could have been encountered Without warning. If the flighterew had desloyed the spoilers at this point without instantly reducing thrust, the control colump pull forces would have increas2d and the speed instability and roll control could have progressed to the point where it would have become impossible to recover the aircraft. Acditionally, if the flighterew sucdenly reduced thrust in an attempt to prevent an overspeed, they could have encountered the pressurization problem that was previously reported end had this distraction to compound their difficulty. Conversely, if ‘he turbulence encounter was such that the aircraft stalled because it crossed its low speed buffet boundary, an uncontrollable wing roli-off and steep tlosedown maneuver could result in a sudden high speed dive. If the flightcrew did not react quickly and appropristely, it elso may have been impossible to recover from such a maneuver. According to the SCR report, the stall speed margins in many of the unmodified wing Model 24 aircraft have been found t) be inadequate, Maintenance of the Stall warning and pusher system is therefore critica: to the safety of flight.

[ANALYSIS] Further calculations, using the radar data, indicated that at the beginning of the final descent, the aircraft experienced an increase in drag, an increase above what is normal in the clean configuration. The reason for this is not known. The linear track of the remaining beacon code positions, as well as the primary radar returns toward the accident site, disclosed a 56° descent angle. Since the ARTCC radar can track an aircraft down to 15,000 feet m.s.i. in the area of the accident site, the Safety Board could not determine the reason radar contact was lost shortly after the aircraft descended below FL 380. Witness observations and a survey of the accident site disclosed that the aircraft struck the ground at a steep angle and at very high speed, resulting in total destruction of the aircraft. Since the aircraft struck the ground at very high speed and the major aircraft structure and flight control surfaces were located in the vicinity of the accident site, the Safety Board concludes that none of the major structures of the aircraft separeted while it was Inflight. The trim position of the horizontal stabilizer actuator jackscrew showed a normal cruise speed trim setting. Analysis of the spoiler actuator indicates that at some point either during the descent or during the impact sequence, the spoilers were in the extended position. The Safety Board believes that, in the absence of conclusive evidence of a mechanical failure or malfunction, an encounter with clear air turbulence of sufficient intensity probably caused the initial altitude upset. Further, the atmospheric conditions could have caused an overspeed, and activation of the stick puller would have resulted in an altitude excursion. It is unlikely that a mistrim condition occurred since the stabilizer actuator jackscrew was in a cruise trim setting, consistent with the radar speed data. An overspeed condition probably would have prompted the flightcrew to extend the spoilers since previous investigations have indicated that extension of the spoilers is a natural reaction to an overspeed. Moreover, this procedure was recommended In earlier AFM's. Extension of the spoilers or the landing gear could be an explanation for the increase in drag as indicated by the radar data. The Safety Board believes that if the crew had lowered the landing gear rather than extending the spoilers, they would have been able to regain control of the aircraft. The aircraft has an adequate range between the onset of high speed buffet and low speed buffet at all altitudes and weight conditions provided there is adherence to the aircraft's performance limitations. Increased load factors caused by manuevering, such as pull-ups or level banked turns, however, will reduce the buffet-free speed range. Additionally, a sharp, unexpected turbulence enzounter can easily cause an aircraft to exceed these margins. Although initial buffet margins can be exceeded, it does not necessarily mean that control difficulties will be immediately encountered. The degree to which the margins are exceeded will determine the aircraft's reaction. Tr - accident aircrafi's buffet margin was 0.14 M,, or about 41 KIAS in 1.5g flight.

[ANALYSIS] Safety Boarc investigators attempted to determine which pilot may have been flying she aircraft at the time of its departure from FL 450. Witnesses were questioned and * recording of the ARTCC tape of communications with the aircraft was played for tlose who knew the flightcrew. However, the physical descriptions of the pilots and thelr positions in the aircraft given by ground service personnel were inconsistent. Additionally, statements regarding which pilot was communicating with Albuquerque ARTCC at 1449:39 were contradictory. Therefore, the Safety Board could not determine who was piloting the aircraft or which seats the pilots occupied at the time of its departure from FL 450, ~19- In view of the total destruction of the aircraft and the leck of CVR and FDR information, the Safety Board was unable to determine precisely the circumstances of the accident. However, the nature of this accident was similar to other Learjet accidents which involved a loss of ccntrol from uigh altitude and from which the flightcrews were unable to recover the aircraft, Accordingly, the analysis of the accident in an attempt to explain how the accident could have occurred is based on the maintenance history, meteorological informetion, ATC radar data, portions of the wreckage, FAA's SC report, and knowledge gained from previous Learjet investigations. 2.2 Airworthiness Between April 1981 and September 16, 1981, the aircraft had been sold four times. There was no record that the successive owners filed maintenance programs at the FAA District Offices having jurisdiction over the areas where the aircraft wav “osed as required by Federal regulation. Since December 2, 1886, a required comprunensive (150-hour) inspection had not been performed by any subsequent owner. Critical items on the aircraft had not been examined closely by qualified maintenance personnel in 10 months, during which time the aircraft had been flown infrequently. Because there was no record, the Safety Board presumes that the open discrepancy concerning the inoperative standby gyro and the lower !atch ef the main cabin door had not been corrected. The Board believes also that a previously reported pressur)xation problem could have been the result of an abnormal leak around the cabin door seal associated with the door latch problem. On December 5, 1980, a previously reported pitch-up problem in the autopilot was corrected and the aircraft was modified in accordance with AD 80-22-10. This modification wes designed to prevent a malfunction in the pitch trim coupler which could also lead to a pitch centrol problem. The April 1981 reported pitch-up problem was apparently corrected through replacement of the AR-1 amplifier module in the pitch synchronization board of the autopilot computer on May 12, 1981, by an authorized Gates Learjet Service Center. However, there were continuous pilot reported discrepancies concerning the autopilot which the Safety Board believes were not associated with the previous maintenance performed. The discrepancies concerned roll oscillation, “wandering” with the heading and altitude hold modes engaged, and the yaw damper’s possible contribution to these control difficulties.

[ANALYSIS] Thus, whe he failed to arrest the decrease by disengaging the PMS and lowering the airplane's nose by rotating the pitch control wheel on the autopilot manual control module, he disconnected the autopilot. As noted earlier, an excursion from & stabslized condition might be exaggerated during the transfer from system monitor mode to system controller because time is needed to ascertain the status of the airplane and assess the situation before the pilut can reenter the control loop and take corrective action. When the autopilot was disengaged, the airplane's excursion fror the stabilized condition was well advaneed and at the point where immediate and proper corrective action was required if the situation was to be remedied safely. The captain was not only unable to assess the situation properly, he was confused by it; therefore, he was unable to take the necessary action to correct the situation. The DFUR data indicated that his actions most probably aggravated the situation. The Safety Board concludes that the captain became spatially disoriented at the onset of the upset and was unable to reorient himself until the airplane bagan to emerge from the clouds. ‘The fact that the first officer was unable to help the captain reorient himself during the descent showed that he also became disoriented during the upset and descent. The Safety Board further notes that the captain did not, as was recommended during his training and in his training manual, disengage the autopilot when the No. 4 engine initially “hung.” Thereafter, he reiied on the autopilot to maintain the alrplane in straight and level flight during the deceleration, and he did not apply left rudder trim to level the control wheel before disengaging the autopilot. Sinee the decreasing airspeed was Initially and readily apparent and would, if allowed to continue unchecked at FL 410, seriously menace the safety of his alrplane, the captain's continuing preoccupation with airspeed control was understandable. However, the captain was an experienced multiengine and Boeing 747 pilot and he also should have known how the loss of thrust _ from an outboard engine would affect an alrplane's controllability, especially when it is coupled with decreasing airspeed. Given his Boeing 747 experience, the captain should have also known that the autopilot's lateral control authority did not include the rudder aiid that the effects of the thrust loss covid only be counteracted by introducing a leftwing-down roll, an aotion which would also introduce a side slip, increase droy, and aggravate the airspeed decrease. Given these circumstances, the Safety Board explored the reasons why the captain was not alert te this condition and why he was not monitoring his attitude direction Indicator more closely during this phase of the operation.

[FINDINGS] The autopilot effectively masked the approaching onset of the loss of control of the airplane. The captain was distracted from his flight monitoring duties by his participation with the flight engineer in the evaluation of the No. 4 engine's malfunction. With the exception of the loss of thrust on the No. 4 engine, no other airplane malfunction effected the performance of the airplane; the loss of thrust on the No, 4 engine did not contribute to the accident. The captain was also distracted by his attempts to arrest the airplane's decreasing airspeed, anc this also contributed to his failure to detect the airplane's increasing bank angle. #17 Safety Recommendation A85-27, Issued LARE ePIC KIS, TLE AED oI 10. 11. ‘he lateral control deflectinns required to maintain level flight under conditions of thrust asymmetry and decreasing alrspeed exceeded the limits of the autopilot's lateral control autnority, causing the airplane to roll and yew to the right. The captain lost control of the airplane when, after disengaging the autopilot, Ke failed to make the proper flight control corrections to recover the airplane. The damage to the airplane was @ result of the acceleration forces and high airspeeds that occurred during the upset and recovery Inaneuvers. 3.2 Probable Cause he National Transportation Safety Board determines that the probable cause of this accident was the captain's preoccupation with an inflight malfunction and his failure to monitor properly the airplane's flight instruments which resulted in his losing control of the airplane. Contributing to the accident was the captain's over-reliance on the autopilot after the loss of thrust on the No. 4 engine. 4. RECOMMENDATIONS None. BY THE NATIONAL TRANSPORTATION SAFETY BOARD March 29, 1986 /s/ JIM BURNETT Chairman /s/ PATRICIA A. GOLDNAN Vice Chairman /s/ JOHN K. LAUBER Member : §. APPENDIXES §.1 APPENDIX A INVESTIGATION AND HEARING 1. Investigation The National Transportation Safety Board was notified of the accident about 1800 eastern standard time on February 19, 1985, and immediately dispatched an investigator from its Los Angeles Field Office to San Francisco. At 0806 eastern standard time, February 20, 1985, an investigative team from Washington, D.C. was dispatched to San Francisco. Investigative groups were established for operat'ons, air traffic control, meteorology, survival factors, airplane structures, airplane syster 3, powerplants, cockpit voice recorder, digital flight data recorder, human performance, and airplane performance. Parties to the investigation were the Federal Aviation Adrainistration, the Boeing Commercial Airplane Company, and the Pratt and Whitney Division of the United Technologies Corporation. The Chinese Civil Aeronautics Administration appointed an accredited representative to assist the Safety Board during the investigation.

[ANALYSIS] Therefore, in its analysis, the Safety Board evaluated data contained in past reports of similar secidents, as well as psychological literature discussing the factors that contribute to breakdowns in decision making and monitoring capability. These areas ‘ineluded boredom, monotonous environmental conditions, fatigue due to circadian desynchronosis, and over-reliance on automated flight systems. In addition, the manner in which the first officer and flight engineer performed during the loss of control sequence was also evaluated in relation to the above areas. Although the first officer was capeble of either flying the airplane or assisting the flight engineer in his analysis of the loss of thrust on the No. 4 engine, the captain did not ‘ask him specifially with either chore. During this period, the additional task levied on the first officer was to obtain clearance from Oakland ARTCC to descend, and the captain did not direct the first officer to obtain an emergency descent clearance. The facts showed that the first officer performed his communications duties in a timely manner; that he had warned the captain of the decreasing airspeed and the increasing right bank; that after the No. 4 engine flamed out he had, without informing the captain, instrueted the relief flight engineer to come forward and help the flight engineer restart the No. 4 engine; and that he came to the captain's assistance on the flight controls without being instructed to do so, Although the first officer was subject to fatigue, boredom, and the same monotonous environment as the rest of the crew, and although he had less off-duty titne during the flight than the captain and flight engineer, he seemed to have performed his assigned duties and overall wonitoring tasks in a timely manner. Given these factors, the Savety Board cannot state with any confidence tat any of the psychological factors that could have reduced his capability to perform affected his actions during the accident sequence. The facts, limited as they are, indicate that his performance was unaffected by these factors. ‘With repard to the captain and flight engineer, both men were performing in @ time spectrum that was later than their typical sleep periods. Although both men had taken a $-hour rest during the flight, the quality of their rest during this period cannot be equated to that which would have been achieved by slea) either at home or in a hotel. Their duty tasks consisted of routine monitoring of the performance of the airplane's automated flight systems, a task that is repetitive and monotonous and capable of producing a state of boredom, The existence of these conditions required the Safety Board to examine the possibility that they might have influenced and derogated the manner In which the flight engineer and captain performed during the emergency. The flight engineer's performance before, during, and efter the loss of control disclosed actions that were correct and timely and other actions that deviated from the required checklist procedure or that demonstrated that he had been unable to analyze correctly the portrayal of the airplane's engine instruments.

[ANALYSIS] The Safety Board concludes that one of the causal factors of *he accident was the captain's reliance on the autopilot while the airplane was decelerating. During this 3 minute 40 second period, the captain allowed himself to remain removed from the "control loop" by leaving the autopilot engaged. As a result, he was net aware of the increasing control inputs required to maintain level flight. Had the captain placed himself in a "hands on" relationship with the airplane by disconnecting the autopilot at .1e onset of the engine problem, he probably would ive been more alert to the increasing asymmetrical forces being exerted on the airplane since he would have been required to make the necessary control inputs to maintain level flight. Since he had no physical relationship with the airplane flight controls, the only cues available to him to monitor the airplane's attitude and performance were the visual cues available from either the airplane instruments or the outside horizon since the airplane was flying above the clouds. However, even under conditions of visual flight, the flight instruments remain the primary tools at high altitudes for maintaining level, stabilized flight in large airplanes. The captain's statement corroborated the fact that he was relying on these instruments for that purpose. Under these conditions, therefore, the primary instrument for attitude control was the attitude director indicator, which may not have concerned the captain initially since it depleted either a wings-level attitude or a very slight left-wing-down bank. With regard to heading, over the period between 1011:09 to about 1014:00, the heading increased about 4°, a change so slight as to be almost imperceptible. Thus, except for airspeed, which concerned the captain greatly, the only thing in the cockpit that would have depicted the worsening control situation was the control wheel's increasing leftwing-down deflection. However, this was an area which was not included in the captain's regular instrument scan pattern, and since he was rot "hands on," he was not aware of the deflection. . Aon en al RNR tse drei ee suet) Wear Lee tee eee A Re SNES gts ETRE Me ect thatthe eeerenem ay Sthemietten peek Ane sani a a He AONE Nee EERE ete eh em BES Ce RPI E A ER EEE p we ree me NT x PP ra Ts ec cace eer ethan ip te AR A OLENA NSPE REO NRI SCA MMO HE CE hme Reg att IN Tn came SIENA 2 Ate NR Cadre AON te! mE dT NAT ETN During the latter part of this perioa, the captain's statement indicated that his attention seemed to be directed almost solely to the airspeed indicator as he tried to arrest the airspeed decrease. Thus, whe he failed to arrest the decrease by disengaging the PMS and lowering the airplane's nose by rotating the pitch control wheel on the autopilot manual control module, he disconnected the autopilot.

[ANALYSIS] The Safety Board also concludes that the captain over-relied on the autopilot and that this was also causal to the accident since the autopilot effectively masked the approaching onset of the loss of control of the airplane. Although tne Safety Board has cited distraction and over-reliasice on the autopilot as causal factors, it also notes that the airplane had been airborne avout 10 hours, that it had traversed several time zones, and that the upset occurred about 0214 Taiwan local time, or about four to five hours after the eaptain had been accustomed to going to sleep. Thus, his ability to obtain, assimilate, and analyze all the data presented to him could have been impaired by the effects of monotony, boredom, and fatigue. However, 1n analysis of the captain's performance does not support a conclusion that the his performance was impaired by these factors. Tne facts and circumstances showed that the captain was alert to the situation as it developed. The data also showed that the captain had five hours rest during the flight, that he had slept two hours during this period, and that he had been at his duty station about 3 hours when the upset occurred. The Safety Board concluded that the preponderance of the evidence showed that the deviations and omissions from prescribed airplane procedures noted in the captain's performanee resulted from the causal factors cited earlier, i.e., distractio. and overreliance on the autopilot. In conclusion, the Safety Board believes that the loss of thrust on the No. 4 engine was the precipitating factor in the accidents however, we do not believe that it should be eesidered a contributory factor. Except on takeoff, nt, or shortly after critical engine failure speed, an engine loss does not require an emergency procedure wherein immediate and memory actions are required of the flightcrew. An engine loss at cruise altitude and at eruise speeds does not place the airplane in immediate jeopardy nor, for the most part, are any immediate responses required of the flightcrew to retrieve the airplane from jeopardy. ‘The facts of this accident confirm this evaluation since the loss of control did not occur until more than 3 minutes after the No. 4 engine had lost thrust. More than enough time was available to the flighterew to react properly and prevent the upset. This fact was amply demonstrated on two previous flights for this airplane in which similar situations occurred; the malfunations were corrected, and the airplane proceeded to scheduled destinations without further incident. The Safety Board is aware of present and proposed Nationa] Aeronautics and Space Administration (NASA) studies on the effects of cireadian desyachrenosis on flighterew performance and efficiency. NASA has recently concluded a study of the effect of circadian desynchronosis on the performance of flightcrews engaged in shorthaul flignts, but has not, to date, releasad its findings.

[ANALYSIS] The Safety Board concluded that a preponderance of the evidence showed that the deviations and omissions noted above resulted from either a lack of knowledge of the airplene systems and procedures, the traurnatic effects of the upset and subsequent descent on the flight engineer's ability to scan his instrument panels, or a combination of these two factors. I . net cen tems! we tant te NRE RTT NE ORME HT ew Kote cen tment ot Ae iat aN gem TG Neh Thm HT In the event of an abnorme! fiight condition, company policy and the AOM dictated that the captain assume control c? che airplane and direct the other crew members to deal with the abnormal condition. Since the captain was at the controls when the flight engineer told him that the No. 4 engine did not accelerate, there was no need for him to take any further action other than to monitor the flight engineer's attempts to analyze the engine's performance and restore it to normal operation. He did not disengage the autopilot since it relegated the tasks involved with flying the airplane to merely monitoring the autopilot's performance. Had he disengaged the autopilot, as recommended in his training, he would have been required to perform the physical, more difficult, and more time and attention consuming tasks involved with flying the airplane manually, I The effects of the asymmetrical thrust condition began to assert themselves at about 1011:10, and the No. 4 engine flamed out about 1012:42. Based on the decrease in pitch attitude and the subsequent momentary airspeed increase, the Safety Board eoneludes that the the PMS was disengaged about 1014:30. Based on the initial movements of the control wheel from its 22.9° left-wing-down position, the Safety Board also concludes that the autopilot was not disengaged until 1014:50. During the 3 minute 40 second period of deceleration, the staternents of the captain and flight engineer showed that the captain was totally cognizant of the engine situation, and thereafter, his attenticn appeared to focus almost exclusively on the airplane's decreasing airspeed. According to the captain, he had disengaged the autopilot in order to lower the nose of the airplane faster and recover airspeed. Although he said that he was aware that the airplane had entered a right bank, he was apparently not aware of the magnitude of the right-wing-down attitude. The Safety Board concludes that one of the causal factors of *he accident was the captain's reliance on the autopilot while the airplane was decelerating. During this 3 minute 40 second period, the captain allowed himself to remain removed from the "control loop" by leaving the autopilot engaged. As a result, he was net aware of the increasing control inputs required to maintain level flight.

[ANALYSIS] Given these circumstances, the Safety Board explored the reasons why the captain was not alert te this condition and why he was not monitoring his attitude direction Indicator more closely during this phase of the operation. Had he done so, he would have noted the airplaie was rolling right-wing-down, that the autopilot could no longer maintain the airplane's heading and roll attitude, and that additional control inputs were required, i.e., rudder or rudder trim. The DFDR readout showed that after the No. 4 engine had "hung," the airplane accelerated to about 250 KIAS and stabilized at that airspeed for about 1 minute 30 seconds. During this period, the autcilot roaintained the airplane at a relatively wings-level attitude with left-wing-down control whee) deflections of about 6° to 10% The full effects of asym metrical thrust were not felt until after the No. 4 engine flamed out. Thereafter, the airplane began to decelerate, its rate of deceleration began to inerease, and the captain's statement showed that his ettention began to focus almost exclusively on the airplane's airspeed. When the captain disconnected the PMS from the autopilot, the airplane was rolling through the 20° right-wing-down attitude and the evidence showed that the captain did not observe the airplane's roll attitude. After disengaging the PMS and inserting a nose-down control correction into the autopilot, the captain continued to monitor the ai*speed indicator to = cassmrtiet wot arenednas Slane ef AH ERNE NOTRE NISSEN elt AGN NO TAMA MLN AOR ARAAESA I) MINI a UE NTS 8 observe the results of the noge-down control correction. During this period, the airplane continued to rok 43 the right and past the 45° right-wing-down attitude. Although the ADI is to the tight of and abuts on the airspeed indieator, the captain never noticed the rightwing-down ADi indications until he disconnected the autopilot. The evidence showed that, starting just before he disconnected the PMS, the captain was distracted by the decreasing airspeed. With the continuing decrease, the captain's distraction with the airspeed increased to the point where his instrument sean patte-n broke down and his visual attention became fixed on the airspeed indicator. The ADI went unobserved. The Safety Board can only conclude that the captain was distracted first by the evaluation of the engine malfunetion and second by his attempts to arrest the decreasing airspeed, and that, because of these distractions, he was unable lo assess properly and promptly the approaching loss of airplane control. The Safety Board also concludes that the captain over-relied on the autopilot and that this was also causal to the accident since the autopilot effectively masked the approaching onset of the loss of control of the airplane.

[ANALYSIS] Although the captain said that the airplane exceeded Vmo twice and also decelerated belew 100 KIAS during the dive, all three crew members said that they did not hear the overspeed warning and that the stall warning stickshaker did not activate. Examination of the reliable recorded airspeed data points showed that the Vmo limitation was not exceeded during the descent. However, the recorder data does show airspeeds at or below 100 KIAS. The Safety Board cannot explain why the stall warning stickshaker did not activate, or if it did activate, why it was not felt or heard by the flightcrew. The Safety Board's investigation and analysis concentrated primarily on two major areas. First, the investigation sought to identify the cause of the loss of thrust on the No. 4 engine, and thereafter to assess whether the actions taken by the flightcrew to cope with the malfunction were reasonable and proper. Second, the invastigation sought to determine why the flighterew was unable to maintain controi of the airplane after the loss of thrust on the No. 4 engine. 21 The Engine Failure About 1010:48, the PMS, in response to the incraased airspeed caused by the wind shear, had decreased the EPRs on all four engines to 0.9 EPR. Then, about 1011;16, i the PMS, in response to the now reduced airspeed, began to advance the four throttles to 74 restore the airplane to the commanded 0.8$M. The investigation of the No. 4 engine and 3 its comporients showed that it had experienced a lean shift of the acceleration schedule ! erie wo + new we “ Pree eines mf * at tare. cet os Beis Mag Bek eee CE ee Ste wa Lars fT Sar ain Bait Soa PASEO NLA eee St OP ra ae ste Ba he ERTS PRET STS SE REE ee ee % Peta resulting in a reduction in the fuel flow available for engine acceleration. A reduction of this type reduces the rate at which the engine would accelerate from flight idle. The DFDR data showed that all four engines started to accelerates however, the dat. viso showed that the No, 4 engine accelerated at a slower rate than the others. As engines Nos. 1, 2, and 3 accelerated, their respective bleed air controllers closed their 15th stage or high stage bleed air valves. Since the No. 4 engine accelerated slower than the other engines, it did not achieve high enough power for its bleed air controller to close the high stage bleed valve at the same time tue high stage bleed air valves w: se closed on the other engines, and the No. 4 engine, at high altitude, probably assumed most of the air conditioning air bleed load. The additional fuel demand imposed by the "bieed load hogging," in combination with the reduced fuel flow available because of the control lean shift, caused the No.4 engine to fail to accejerate and to "hang" at slightly above 1.0 EPR.

[CONCLUSIONS] The captain did not respond immediately to the stick shaker warning, which was followed within 2 seconds by a loss of Jateral control. Thereafter, the continued exertion of back force on the control column was inappropriate. The airplane recovered from the out-of-control descent when control forces were relaxed and the landing gear was lowered. The operation of the engines and propellers was normal until after the loss of control. The captain shut down the left engine and feathered the propellers, mistakenly believing that there was an engine overspeed. Three of the four left propeller blades and the cowlings separated after the beginning of the event, during the post-stall gyration. 31 15. Following recovery, due to asymmetric aerodynamic drag caused by the damaged engine, propeller, and cowl, the airplane was unable to maintain level flight, and precise airplane control was not possible. 16. Because of the inability to precisely control the airplane after the recovery, the flightcrew landed long. This, and the fact that the runway was wet, precipitated the overrun landing roll, subsequent airplane damage, and injuries. 17. The crew rest periods scheduled for the trip sequence were within company guidelines and FARs. However, the crew did not take advantage of the rest periods, and the combined effects of cumulatively limited sleep, a demanding day of flying, and a time of sn, day associated with fatigue were factors in the crew’s inadequate judgment and performance. 18. Although coordination between Little Rock Approach Control and Memphis ARTCC could have been much improved, it did not contribute to the accident. ‘ 12. PROBABLE CAUSE The National Transportation Safety Board determines that the probable causes of this accident were the captain’s failure to maintain professional cockpit discipline, his consequent inattention to flight instruments and ice accretion, and his selection of an improper autoflight vertical mode, all of which led to an aerodynamic stall, loss of control, and a forced landing. Factors contributing to the accident were poor crew discipline, including flighterew coordination before the stall and the flightcrew’s inappropriate actions to recover from the loss of control. Also contributing to the accident was fatigue induced by the flightcrew’s failure to properly manage provided rest periods. 13. RECOMMENDATIONS As a result of its investigation, the National Transportation Safety Board recommends that the Federal Aviation Administration: 32 Require that 14 CFR Part 135 air carriers provide aircrews, as part q\ f of their initial and recurrent training, information on fatigue i countermeasures relevant to the duty/rest schedules being flown by the company. (Class II, Priority Action) (A-94-73) BY THE NATIONAL TRANSPORTATION SAFETY BOARD Carl W. Vogt Chairman Susan M. Coughlin Vice Chairman John K. Lauber Member John A. Hammerschmidt Member James E.

[CONCLUSIONS] In spite of these anomalies, the Board believes that the failure to pass on information about the ILS outage and about men and equipment on the runway did not contribute to the accident sequence of events. The weather in the Pine Bluff local area was VFR, and by the time that the aircraft had landed, the full length of the runway was available. 11. CONCLUSIONS 1. The airplane was properly certificated, equipped, and maintained in accordance with FAA regulations and approved Continental Express procedures, 2. The airplane was dispatched in accordance with FAA regulations and operator procedures and was within the prescribed limits for weight and center of gravity. 3. The Continental Express operations and maintenance procedures pertinent to the conduct of the accident flight were found to be logical, clearly presented, and in accordance with the FARs. In many instances, the operator’s procedures and requirements exceeded the minimum standards set by the FAA. 4. Both crewmembers were properly certificated, qualified, and current for the operation being conducted. 5. All of the flight control, autoflight, stall warning, and flight instrument systems were operating normally up to the time of the loss of control. No evidence of primary or trim flight control system malfunction was found. 6. The freezing level was near 11,500 feet and the potential for icing existed up through 19,000 feet. The airplane was in clouds with zero visibility, and the tops of the clouds extended above 21,000 feet. 10. 11. 12. 13. 14. 30 The entire crew violated the sterile cockpit rule as the airplane was passing through 8,000 feet. In addition, the flight attendant was present in the cockpit as the airplane climbed above 10,000 feet, engaging in nonpertinent conversation with the captain, for 4 minutes and 27 seconds up to and during the loss of the control. The captain and the first officer failed to adequately monitor the progress of the flight during the climb, and the first officer failed to adequately monitor the captain’s actions. The captain engaged the autoflight system in the "heading" and "pitch hold" modes during the climb, obviating the stall and speed protection afforded by the other vertical modes. This autoflight system configuration was contrary to the company’s training and procedures. During the climb, the pitch was increased by the captain, using the autoflight “pitch hold" mode, in the minutes before the loss of control. LS The increase in pitch, and subsequent loss of airspeed, resulted in an aerodynamic stall. The stall and loss of contro! at a higher-thanexpected airspeed was caused by aerodynamic performance degradation due to wing ice contamination. The captain did not respond immediately to the stick shaker warning, which was followed within 2 seconds by a loss of Jateral control. Thereafter, the continued exertion of back force on the control column was inappropriate. The airplane recovered from the out-of-control descent when control forces were relaxed and the landing gear was lowered. The operation of the engines and propellers was normal until after the loss of control.

[ANALYSIS] It appears to the Safety Board that the basic stability of the airplane when the control column was finally returned to neutral and perhaps the lowering of the landing gear were the only factors that prevented an uncontrolled descent into the terrain. The Safety Board believes that this accident illustrates the need to emphasize to pilots the aerodynamic fundamentals of a stall-induced loss of control and the need to move the control column to reduce the angle of attack to recover from such a loss of control. Flightcrew Fatigue The accident flight came at the end of the crew’s 3-day flight schedule. The first day of the schedule was demanding and culminated in a reduced rest period. The second day was short, with the crew going off duty about 1130 and not having to report back until 0530 the next day. The last day was perceived by the crew as being the most demanding because it was the end of the trip, and as the first officer said, "one is just ready to go home and see the family." The captain stated that the workload was slightly heavier on the last day due to having seven legs to fly in IMC. : The crew rest periods scheduled for the trip were within company guidelines and FARs. The crewmembers had sufficient opportunity on the second day of their flight schedule to get adequate rest; however, they did not take advantage of this opportunity. For the two nights before the accident, the pilots averaged only about 5 to 5 1/2 hours of sleep per night. The accident occurred after a long and relatively difficult day of flying and on the last leg when the crew anticipated getting home. Further, the accident occurred in the late afternoon when the human body normally reaches a physiological low level of performance and alertness. The Safety Board believes that the combined effects of cumulatively limited sleep, a demanding day of flying, and a time of day associated with fatigue had an effect on crew performance. The Safety Board recently examined the 37 major air carrier accidents from 1978 through 1990 for which human performance issues were cited in the probable cause determination ("A Review of Flightcrew-Involved, Major Accidents of U.S. Carriers, 1978 through 1990." Safety Study NTSB/SS-94/01). Many human 28 performance background variables were compared to the types of errors observed in the accident sequence in an effort to identify factors that might be useful in accident prevention. Several fatigue-related variables were examined--time since awakening, time of day, time zone crossings, and changing work schedules. It was found that the time since awakening for each pilot related to significant differences in performance, in terms of the number and types of errors made by pilots.

[ANALYSIS] The Safety Board considered and could not conclusively discard the possibility that the nacelle damage was caused by the maneuvering and air loads imposed on the structure during the out-of-control descent. If this were the case, the distortion of the nacelle and engine mounts could have moved the propeller control linkage to feather, which in turn would have resulted in a less of blade centrifugal loads, which permitted the blades to separate. The Safety Board believes that it is more probable that the loss of the propeller blades was initiated when the crew attempted to feather the left engine in the belief that an engine overspeed had occurred. The loss of centrifugal loads on the blades as the propeller rotational speed slowed, combined with the severe cyclic loads imposed on the blades-during the departure from stabilized flight, would have allowed the blades to rock in the hub with consequent failure of the blade retaining rings, release of the bearings, and blade separation. The Safety Board believes that the imbalance of the propeller assembly as the blades separated imposed the high vibratory loads to the engine shaft, which produced the nacelle damage. Thus, the Safety Board believes that while the propeller blade separation was a result of the loss of control, it contributed to the severity of the accident, in that the effective loss of one engine and the nacelle damage resulted in a performance degradation that limited the captain’s ability to control the airplane and maintain altitude. The limited control led to a long landing touchdown and subsequent overrun. Flightcrew Performance Selection of Improper Autoflight Mode The Safety Board believes that the guidance provided in Continental Express’ manuals and training programs is clear in the description of the autoflight system operating modes. These materials specifically state that climbs should be conducted in either the "climb" or "airspeed" modes to ensure that an adequate airspeed margin above stall is maintained. Contrary to that guidance, the captain chose to use the “pitch hold" mode. Moreover, the captain increased the pitch attitude to increase the rate of climb apparently without reference or concern about the performance capability of the airplane to maintain a safe airspeed. Consequently, the airspeed decreased, and as a result of increased drag, the rate of climb decreased. That the captain’s action to select the “pitch hold" mode was intentional is evident from his postaccident statements and his attempt to increase the pitch attitude in response to the flight attendant’s request to enhance the rate of climb. The Safety Board is particularly concerned that an experienced pilot with over 2,500 hours in the airplane type would fail to recognize the loss of efficiency as well as the potential danger in selecting a constant pitch attitude climb. The Safety Board believes that the captain’s action was not only contrary to company procedures, but contrary to the principles of basic airmanship. Of equal concern is the inaction of the first officer to question the captain or monitor the autoflight system selection. The Safety Board concludes that the captain’s inappropriate selection of the "pitch hold" mode combined with the flightcrew’s subsequent failure to maintain a safe airspeed was the primary cause of this accident.

[ANALYSIS] Stall Recovery The procedures for stall recovery are delineated in the Continental Express Operations Manual, and pilots are required to demonstrate their knowledge of these procedures repeatedly during training. However, the Safety Board believes that the pilot training for stall recovery is directed toward low altitude encounters 26 where minimum altitude loss is critical. Further, during training, the pilot is expected to respond immediately to the first activation of the stick shaker where sufficient margin from full stall or loss of control exists and aggressive action to reduce pitch attitude is unnecessary and in fact would not be consistent with a minimum altitude loss recovery. Therefore, while the operations manual also states that it may be necessary to lower pitch attitude to trade altitude for airspeed if an impending stall is encountered at cruising altitude, and while this procedure is stressed in the most basic pilot training courses, it might not be an immediate or reflexive response to stick shaker onset. Also, the approach to stall demonstrations during training are conducted with an airplane or simulator having normal aerodynamic performance characteristics, that is, there is no consideration given to the performance degradation or the effect on stall warning system margin that result from ice accretion on the wing leading edge. During this accident, ice accretion on the wing significantly reduced the margin between the stick shaker onset and the loss of control. The FDR and CVR correlation show that within 2 seconds of stick shaker onset and autopilot disconnect, the airplane entered a sudden and uncontrollable roll oscillation. The data then show that instead of relaxing control column force, the captain increased back force to hold the control column aft and introduced roll commands through the control wheel that were initially out of phase with the proper corrective deflections. Thus, the captain’s initial control deflections following the stick shaker onset and-the almost immediate loss of control aggravated, rather than corrected, the out-of-control maneuvers. The FDR data indicated that the airplane was capable of developing a positive load factor throughout the uncontrolled decent. Thus, the Safety Board believes that recovery could have been accomplished with minimum altitude loss at the time of the stick shaker activation or earlier in the descent had the captain relaxed control column force. Had he done so, the angle of attack would have been reduced and aileron effectiveness restored permitting the airplane to regain wings-level flight. The Safety Board acknowledges, however, that any crew, regardless of their training in the recovery from unusual attitudes, may have had difficulty responding to a situation such as that confronting this crew after the lateral control loss--that is, high lateral and vertical acceleration loads that combined with a lack of visual reference and the rapidly changing attitude 27 instruments with corresponding changes in altimeter and airspeed readings would have produced disorientation. It appears to the Safety Board that the basic stability of the airplane when the control column was finally returned to neutral and perhaps the lowering of the landing gear were the only factors that prevented an uncontrolled descent into the terrain.

[ANALYSIS] The Safety Board believes that the captain’s action was not only contrary to company procedures, but contrary to the principles of basic airmanship. Of equal concern is the inaction of the first officer to question the captain or monitor the autoflight system selection. The Safety Board concludes that the captain’s inappropriate selection of the "pitch hold" mode combined with the flightcrew’s subsequent failure to maintain a safe airspeed was the primary cause of this accident. Flightcrew Inattentiveness The recorded cockpit conversation between the captain, first officer, and flight attendant was consistent with a complacent and lax atmosphere throughout the flight. Having selected the "pitch hold" mode for the climb, it was particularly important that the flightcrew monitor the essential flight instruments continually to maintain a safe airspeed and positive rate of climb. Instead, the captain permitted the flight attendant to enter the cockpit and then engaged in casual 25 conversation for over 4 minutes before the loss of control occurred. Meanwhile, the first officer was making entries into the airplane’s log book, which diverted his attention from the flight instruments. It was in response to the flight attendant’s request that the captain selected an increased pitch attitude. Subsequently, he continued to talk to the flight attendant and was not attentive to his flight instruments. The Safety Board believes it likely that the captain began to dial in nose-right rudder trim as a normal action during climb and, without observing the turn-and-slip indicator, continued to do so until full trim was reached. A passenger observed the captain "turning a knob" that was located in the approximate position of the rudder trim knob. The investigation determined that there were no autoflight or trim system malfunctions that could have resulted in full nose-right rudder trim. Further, neither of the flightcrew remembered observing the wing leading edges or propeller spinners to check for ice accretion even though they were flying through visible moisture and freezing temperatures. The Safety Board believes that the flightcrew’s inattention to the flight led directly to their failure to maintain a safe airspeed. ‘ It is probable that, during their last sequence of flights together, all three crewmembers became too relaxed with each other and, consequently, less professional in their relationship during flight. For instance, the open cockpit door and the non-use of headsets by the flightcrew encouraged and certainly allowed the flight attendant to distract the pilots for several minutes while a critical, unsafe flight situation developed. The crewmembers were apparently comfortable enough together to allow themselves to become extremely complacent, and the lax cockpit atmosphere set by the captain was accepted by the other crewmembers. All three individuals should have done more to prevent the accident situation from developing, and good crew coordination and resource management principles certainly would have assisted them. Stall Recovery The procedures for stall recovery are delineated in the Continental Express Operations Manual, and pilots are required to demonstrate their knowledge of these procedures repeatedly during training. However, the Safety Board believes that the pilot training for stall recovery is directed toward low altitude encounters 26 where minimum altitude loss is critical.

[ANALYSIS] Roll oscillations as high as 90° in each direction and pitch attitudes as low as 67° airplane nose down were recorded during the descent. Coincident with the roll oscillations, the airspeed reached about 210 KIAS, and the airplane, while remaining near a stall condition, developed a positive load factor between 2 and 3 Gs. The FDR indicated control wheel movements left and right initially out of phase with the airplane’s roll oscillations. The control column was moved aft to command airplane nose up throughout the descent. According to the flightcrew, the recovery was initiated soon after the first officer lowered the landing gear. However, the FDR showed that the control column and wheel were returned to near neutral about the same time. FDR data show that the minimum altitude during the loss of control was approximately 5,600 feet. The airplane then climbed rapidly and briefly entered a second stall at about 6,700 feet and ultimately returned to controlled flight at 5,500 feet. The loss of control, descent, and recovery all occurred in IMC. Comparison of Theoretical and Actual Airplane Performance The crew stated that the autoflight system was configured in the pitch and heading hold modes throughout the climb. A review of the manufacturer’s performance data for the gross weight and atmospheric conditions that existed indicated that for an international standard atmosphere (ISA) +0° day, at an altitude of 17,000 feet, climb speed should have been 155 KIAS, which would have resulted in a climb rate of 1,333 fpm. In ISA +10° conditions, the target climb speed would have remained the same, and the rate of climb would have been 800 fpm. The performance data also indicated that for the conditions present, the stick shaker should have activated at 127 KIAS, and the airplane would have stalled at 117 KIAS. The stick shaker on the accident airplane activated at 141 knots, or 14 knots higher than expected. The theoretical angle of attack at which the stick shaker activates is 10°. Although angle of attack is not recorded directly on the FDR, a comparison of recorded pitch attitude and computed flightpath angle confirmed that the actual angle of attack was about 10° when the stick shaker activated. 21 Although the deceleration of the airplane below the normal climb speed was a direct result of the captain’s selection of a higher-than-normal pitch attitude, the activation of the stick shaker and the loss of lateral control at airspeeds 14 and 22 knots higher than the theoretical speeds for those events indicated that the aerodynamic performance of the airplane was affected by still other factors. Two of these factors were examined on the EMB-120 engineering simulator: drag due to sideslip and ice accretion.

[ANALYSIS] The passenger’s observation of a whitish substance on the windshield, which appeared to be snow, would be consistent with some amount of ice accretion. The Safety Board believes that an accretion of ice on the wing is the only reasonable explanation for the occurrence of stick shaker activation and loss of roll control at higher-than-expected airspeeds. The Safety Board believes that only a small amount of ice on the wing’s leading edge could have had a significant effect on the aerodynamic performance under the circumstances of this flight. If the airplane accumulated ice during the climb above 11,500 feet while at a relatively low angle of attack, the ice would have formed at the stagnation point® associated with that angle of attack. As the airplane slowed, the corresponding inorease in angle of attack would have resulted in a movement of the stagnation point lower on the leading edge. Thus, the ice that had formed at the higher speed would be above the new stagnation point and produce a greater disruption of the air flow over the wing upper surface leading to premature boundary layer’ separation. The result would be a progressive reduction in the lift produced by the wing and a stall at a lower angle of attack. In past aviation accident investigations, the Safety Board has determined that almost imperceptible amounts of ice, 1/4 of an inch or less, on the wing leading edge has significantly increased the stall speed and lateral control capability of the airplane. Thus, the Safety Board believes that ice accumulated during the climb and resulted in a stall at higher-than-normal speed. While it is likely that the accretion of ice alone would not have led to a stall had the captain attempted to maintain a The stagnation point is the point on the leading edge of the airfoil where the relative airflow diverges to pass above and below the wing so that the local airflow velocity is zero. ?The boundary layer is the airflow immediately adjacent to the wing surface. 23 target airspeed instead of a target pitch attitude, the Safety Board cites the captain’s inattention to ice accretion as a factor in the accident. Propeller Failure Evaluation of the FDR, FDAU, and CVR data revealed that operation of the engines and propellers was normal until after the loss of control occurred. This information corresponded with the crew’s statements. Evidence obtained after the accident from the left engine indicated that at some point the engine had been operated under extremely high vibratory loads at low engine speeds. The FDR revealed that the right engine and propeller continued to operate within normal parameters until the collision with the terrain as the airplane departed the runway during the landing. The Safety Board considered and could not conclusively discard the possibility that the nacelle damage was caused by the maneuvering and air loads imposed on the structure during the out-of-control descent. If this were the case, the distortion of the nacelle and engine mounts could have moved the propeller control linkage to feather, which in turn would have resulted in a less of blade centrifugal loads, which permitted the blades to separate.

[ANALYSIS] Two of these factors were examined on the EMB-120 engineering simulator: drag due to sideslip and ice accretion. Drag due to Sideslip The captain’s observation that the ball in the turn-and-slip indicator was slewed full left and that the rudder trim wheel was trimmed 10 units right immediately before the loss of control indicated that the airplane was in a left sideslip as it approached the stall. The FDR lateral acceleration and control wheel position parameters show that a left sideslip developed about 20 seconds before the autopilot disconnect. However, as the airplane slowed during the climb, right rudder trim would have been required to balance the normal left turning forces produced by propeller effects at high angles of attack. At 140 KIAS, the airplane would have required about four units of nose-right rudder trim. Thus, the sideslip resulting from full nose-right trim was less significant than it would have been at a higher climb speed. When examined in the engineering simulator, it was evident that the reduction in climb performance resulting from the sideslip drag was less than that encountered on the accident flight. Furthermore, because of the short time of the sideslip condition, the Safety Board believes that the out-of-trim condition was not a factor in the loss of control. Effect of Ice Accretion The meteorological data and the observation of other pilots indicate that the conditions present at the time of the accident were conducive to the accretion of ice on the airplane’s aerodynamic surfaces, which would have affected performance. The airplane was in clouds as it climbed above the freezing level at 11,500 feet and was exposed to freezing temperatures and visible moisture for over 7 minutes before the loss of control occurred. While the effect of ice on aerodynamic performance is well known, the ability to quantify the effect in terms of the lift decrease and drag increase associated with specific amounts of ice is 22 limited. Embraer had aerodynamic performance data available for a wing having an inch or more of rough rime ice on the leading edge. When these data were examined in the simulator, the noted degradation in the airplane’s climb performance was far greater than the degradation evident for flight 2733. Although the captain and first officer both stated that they had not observed ice on the wings, there is no evidence that they looked for ice at any time during the climb above freezing level. The passenger’s observation of a whitish substance on the windshield, which appeared to be snow, would be consistent with some amount of ice accretion. The Safety Board believes that an accretion of ice on the wing is the only reasonable explanation for the occurrence of stick shaker activation and loss of roll control at higher-than-expected airspeeds.

[CONCLUSIONS > FINDINGS] Consequently, the Safety Board believes that the FAA should modify the operating and airworthiness regulations of Title 14 CFR or issue appropriate guidance material to clarify airworthiness and operational procedural requirements for conducting FEFs in transport-category aircraft. 48 3. CONCLUSIONS 3.1 Findings 1. The pilot flying made a timely decision to terminate the clean stall and begin the stall recovery. 2. The pilot flying applied inappropriate control column back pressure during the stall recovery attempt in an inadequate performance of the stall recovery procedure established in ABX’s operations manual. 3. Aircraft weight and balance were not a factor in the greater-thanexpected buffet onset speed; and the airplane was loaded within its approved weight and balance limits during the accident flight. 4. Some combination of airframe icing, flight control rigging, or other factors resulted in the greater-than-expected buffet onset speed; however, any effects of airframe icing or flight control rigging upon the stall speed of the accident airplane were minimal. 5. Although the pilot flying trimmed the airplane below the recommended minimum trim speed for the clean stall, this action did not contribute to the accident. 6. The pilot not flying, in the right seat, was serving as the pilot-incommand of the accident flight and was conducting instruction in functional evaluation flight procedures; and the pilot flying, in the left seat, was serving as the second-in-command and was receiving instruction in functional evaluation flight maneuvers, including the clean stall maneuver. 7. The pilot not flying, as the pilot-in-command, failed to recognize, address and correct the pilot flying’s inappropriate control inputs. 8. The absence of the stick shaker prior to the stall did not affect the flightcrew’s recognition of the initial entry into the stall. 9. The inoperative stall warning system failed to reinforce to the flightcrew the indications that the airplane was in a full stall during the recovery attempt. 10. This accident might have been prevented if the flightcrew had been provided a clear, direct indication of the airplane’s angle of attack. 49 11. The engine compressor surges in the No. 2 engine (caused by airflow disruption) may have distracted the flightcrew during the critical early period of the stall recovery, when sufficient lateral control was available for the recovery. 12. The flightcrew did not have a clearly visible natural horizon because of darkness and clouds above and below the airplane, and the airplane most likely encountered instrument meteorological conditions soon after descending through 13,500 feet and remained in instrument meteorological conditions until just before impacting terrain. 13. By conducting these maneuvers without a visible natural horizon, the flightcrew was deprived of an important flight attitude reference that would have aided in their recovery from a full stall. 14. The flightcrew’s exposure to a low fidelity reproduction of the DC8’s stall characteristics in the ABX DC-8 flight training simulator was a factor in the pilot flying holding aft (stall-inducing) control column inputs when the airplane began to pitch down and roll. 15.

[ANALYSIS] Evaluation criteria for the performance of this maneuver by line flightcrews during proficiency checks include minimum altitude loss and avoidance of a secondary stall (which can be recognized by reactivation of the stick shaker or aerodynamic buffet). 31FDR data on control column position indicated that the elevator controls were not jammed during the loss of control sequence. There was forward and aft movement of the column, but the column was moved forward of the 10-degree nose-up position only once, momentarily. 32 The flightcrew had initiated the clean stall maneuver at 13,500 feet, within a block altitude range of 13,000 feet to 15,000 feet. Their use of the lowest 500 feet of the block altitude indicated that the flightcrew anticipated recovering from the stall with a minimum altitude loss, just as they were accustomed to performing and instructing the standard ABX stall maneuvers in the simulator. Further, their execution of the clean stall only slightly above the cloud tops suggests that the flightcrew did not anticipate the possibility of greater altitude loss. Pilots control airplanes, including in stall recoveries, by first establishing appropriate pitch attitudes and power settings, then monitoring the effects of pitch and power on the performance instruments, such as the airspeed and vertical speed indicators. Although the PF attempted to establish his desired pitch attitude and power setting, he failed to recognize that these initial pitch and power inputs were inadequate for the stall recovery he was executing, and he failed to take further action to correct for the decreasing airspeed and developing sink rate. Allowing the control column to move forward would have stopped the airspeed loss, and the airplane would have recovered from the stall. However, he failed to do so. The Safety Board concludes that the PF applied inappropriate control column back pressure during the stall recovery attempt in an inadequate performance of the stall recovery procedure established in ABX’s operations manual, and these control inputs were causal to the accident. Further, although the stall recovery procedure established by ABX cautioned pilots to relax enough control column back pressure to avoid secondary stalls during the recovery, the Safety Board notes that the objective of a minimum altitude loss procedure, including the ABX procedure, is to use the minimum reduction in pitch attitude required to recover from the stall. As such, the minimum altitude loss stall recovery procedure places the airplane at a more critical angle of attack and drag configuration during the initial recovery from the stall, compared to an alternative procedure that uses a greater pitch reduction and trades a greater altitude loss for a more rapid reduction in angle of attack and increase in airspeed. ABX’s role in specifying the type of stall recovery procedure to be used during FEFs is discussed in section 2.5.1. 2.2.2 Buffet Onset and Stall Speeds Prior to executing the clean stall maneuver, the flightcrew calculated the airspeed values at which the stall warning stick shaker should activate and the airplane should stall, based on their estimates of the airplane’s weight at the time the maneuvers began.

[ANALYSIS] Further, because their experience with stalls in the DC-8 was obtained in a simulator without a stall break, the PF and PNF could not practice the nose-down control inputs required to recover a stalled airplane that is pitching down or at a nose-low attitude. Moreover, because the PF and PNF were exposed during extensive simulator experience to what they presumed was the stall behavior of the DC-8, the stall break that occurred in the airplane most likely surprised them. The Safety Board concludes that the flightcrew’s exposure to a low fidelity reproduction of the DC-8’s stall characteristics in the ABX DC-8 flight training simulator was a factor in the PF holding aft (stall-inducing) control column inputs when the airplane began to pitch down and roll, which contributed to the accident. The Safety Board has previously expressed concerns about the inadequate fidelity of air carrier pilot training simulators and their deficiencies in reproducing handling characteristics of an airplane during specific maneuvers.38 The FAA does not require air carrier flightcrews, for example, to be trained in full or deep stall maneuvers, and simulators are not required to be programmed to provide fidelity farther into the stall than the initial buffet or stick shaker.39 Consistent with FAA requirements, simulator manufacturers do not routinely obtain accurate data about airplane stall characteristics from airplane manufacturers.40 However, the Safety Board is aware that for many airplane types (including the DC-8), data obtained from stall maneuvers performed during the certification process could be used to improve the fidelity of stall characteristics in air carrier flight simulators. Further, the stall phase of the flight envelope is one that has received increasing attention in air carrier pilot training through the advent of “advanced maneuver” or “selected event” training programs.41 38See Aircraft Accident Report—“Runway Departure During Attempted Takeoff, Tower Air flight 41, Boeing 747136, N605FF, JFK International Airport, New York, December 20, 1995” (NTSB/AAR-96/04); and Aircraft Accident Report—“Uncontrolled Collision with Terrain, Air Transport International, Douglas DC-8-63, N782AL, Kansas City International Airport, Kansas City, Missouri, February 16, 1995” (NTSB/AAR-95/06). 39See FAA Advisory Circular 120-40B, “Airplane Simulator Qualification.” 40However, regardless of the level of simulator approval, a flight training simulator is required to provide “stall buffet to, but not necessarily beyond the FAA certified stall speed, Vs.” 41See section 1.18.6. 41 Improved simulator fidelity may therefore benefit all pilots and passengers.

[ANALYSIS] Based on the thrust application and the “Set max power” statement of the PF, the Safety Board concludes that the PF made a timely decision to terminate the clean stall and begin the stall recovery. For 8 seconds following the initiation of the stall recovery, from 1808:13 through 1808:21, the PF maintained the airplane’s pitch attitude at between 10 degrees and 14 degrees ANU. During this 8-second period, the airspeed continued to decrease and the airplane entered a fully developed stall. The failure of the airplane to recover before entering the full stall resulted from the control column inputs the PF was making to maintain pitch attitude. The control column was moved aft by the PF, from 5 degrees aft (at 1808:11, just prior to initiating the recovery) to 20 degrees aft (14 seconds later). An increasingly downward flightpath angle coupled with a relatively constant pitch angle resulted in an increasing angle of attack. The increase in angle of attack, which placed the airplane farther into the stalled condition, may not have been perceived by the flightcrew unless they were closely monitoring the airspeed indicator. In addition, the vertical speed indicator and altimeter should have provided evidence of a developing sink rate and stall. Thereafter, the airplane began a series of roll reversals, and the airplane remained in an aerodynamic stall condition because the PF held significant back pressure on the control column all the way to impact. Each time the airplane developed a large nose-down pitch rate (combined with reductions in airspeed at 1808:25 and 1809:22), the PF responded with additional back pressure, according to FDR data on control column movement. In contrast, the appropriate pilot response to an uncommanded decrease in pitch attitude (which is, itself, an indication that the airplane is in a stall) would have been forward movement of the control column.31 ABX’s stall recovery procedures, established in accordance with FAA standards, were used to train line flightcrews to recognize an impending stall and perform a recovery with a minimum loss of altitude. The training procedures call for the airplane to be slowly decelerated at a constant altitude until stall recognition; i.e., the onset of a pre-stall buffet or activation of the stick shaker. Recovery is executed by adding power, reducing pitch attitude slightly (until the stick shaker stops) and maintaining heading. Evaluation criteria for the performance of this maneuver by line flightcrews during proficiency checks include minimum altitude loss and avoidance of a secondary stall (which can be recognized by reactivation of the stick shaker or aerodynamic buffet). 31FDR data on control column position indicated that the elevator controls were not jammed during the loss of control sequence.

[ANALYSIS] 2. ANALYSIS 2.1 General The three-member flightcrew was properly certificated and qualified in accordance with applicable Federal regulations and company requirements. There was no evidence that any medical, behavioral, or physiological factor affected the flightcrew on the day of the accident. Both pilots were qualified in accordance with Federal regulations and company requirements to act as DC-8 PICs and to fulfill the duties of the roles they were assigned, although neither the FAA nor ABX had established any additional experience requirements or qualifications for flightcrew members to perform FEFs (see section 2.5.3 for a discussion of this issue). The airplane was properly certificated, equipped and maintained in accordance with Federal regulations and approved procedures. There were no open or deferred maintenance items listed on the airplane before the accident flight, and there was no evidence that failures of the airplane structures, flight control systems or engines contributed to the accident. Other than the failure of the stall warning system stick shaker to activate (which is addressed in section 2.3.1), there was no evidence of mechanical malfunctions. All records for airplane systems indicated proper maintenance and normal operation. Airplane logs indicated that all four engines had received approved overhaul and noise attenuation modifications. There were no open or deferred maintenance items involving the engines and all inspections were conducted within required intervals. The Safety Board’s analysis of the accident focused on flightcrew performance during the accident sequence, and whether the cues presented to the flightcrew, their prior experience, the fidelity of the ABX DC-8 flight training simulator, the organizational structure and function of ABX, or oversight by the FAA may have affected the flightcrew’s performance. 2.2 Flightcrew Performance During the Accident Sequence 2.2.1 Execution of the Stall Recovery The purposes of the FEF stall series were to verify that the airspeeds at which the airplane experienced stick shaker activation and stall indication were in accordance with precalculated values for the DC-8-63 and to determine whether maintenance performed had changed any of the airplane’s flight characteristics. A comparison between the CVR transcript and the ABX procedures for executing a clean stall (see section 1.18.3) indicates that the flightcrew prepared for the stall maneuver generally in accordance with ABX procedures, with the exception of their use of pitch trim, which is discussed in section 2.2.3. The PF reduced the airspeed into the stall region at approximately the desired rate of 1 knot per second, and he increased engine EPR (spooled up the engines) prior to the onset of buffet to provide adequate engine acceleration during the recovery. 31 By 1808:13, 2 seconds after the flight engineer stated, “That’s the stall right there * ain’t no shaker,” significant engine thrust had been applied, and the PF had stated, “Set max power.” The flightcrew had recognized the incipient stall and initiated the stall recovery in accordance with ABX’s procedures.

[ANALYSIS] FDR data on the performance of the airplane during its takeoff on the accident flight were consistent with the flightcrew’s calculations of the airplane’s weight and stall speed. Therefore, the Safety Board concludes that aircraft weight and balance were not a factor in the greater-than-expected buffet onset speed. Further, based on the Safety Board’s estimates of weight and balance for the accident flight, which were developed using ABX records of the basic operating weight of N827AX and the fuel weight remaining after the December 21 FEF, the Safety Board concludes that the airplane was loaded within its approved weight and balance limits during the accident flight. The flight had been operating at least intermittently in the cloud tops and belowfreezing air temperatures, where weather conditions were conducive to light-to-moderate icing for a brief period before the attempted stall maneuver. The flightcrew’s statements on the CVR, at 1748:34, indicate that the airplane was accumulating airframe ice. Recorded comments of the flightcrew further indicated that the airplane departed from the icing conditions shortly thereafter. The Safety Board was unable to determine the amount of airframe ice that the airplane would have accumulated during this period. However, airframe icing could have caused the airplane to buffet at a greater airspeed than the flightcrew expected. The buffet and stall speeds also could have been affected by the rigging of the airplane’s flap and aileron control surfaces. These control surfaces had been re-rigged prior to the FEFs of December 21 and 22, 1996, as a routine part of the overhaul of N827AX. One of the purposes of performing the stall series during the post-modification FEF was to verify that control surface rigging was proper by comparing calculated stick shaker activation and stall speeds to the airspeeds at which the airplane actually encountered these events. 32Stall speed is an increasing function of an airplane’s weight. An accumulation of ice on the lift-producing surfaces of an airplane can increase the stall speed by altering the shape of these surfaces. A control surface that is rigged at or beyond the forward (retracted) tolerance limit decreases the airplane’s wing area and the camber (curvature) of the wing, increasing the stall speed. Also possibly affecting the stall speed are load factor (vertical G), aileron deflection, spoiler deflection, sideslip, and smoothness of the wing leading edge. 34 Consequently, like airframe icing, variations in flap and aileron rigging could have caused the airplane to buffet at a greater airspeed than expected. The Safety Board was unable to determine the extent to which these possible conditions individually contributed to the early onset of buffet, but concludes that some combination of airframe icing, flight control rigging, or other factors resulted in the greater-than-expected buffet onset speed.

[CONCLUSIONS > FINDINGS] By conducting these maneuvers without a visible natural horizon, the flightcrew was deprived of an important flight attitude reference that would have aided in their recovery from a full stall. 14. The flightcrew’s exposure to a low fidelity reproduction of the DC8’s stall characteristics in the ABX DC-8 flight training simulator was a factor in the pilot flying holding aft (stall-inducing) control column inputs when the airplane began to pitch down and roll. 15. The accident could have been prevented if ABX had institutionalized and the flightcrew had used the revised functional evaluation flight stall recovery procedure agreed upon by ABX in 1991. 16. ABX’s failure to require completion of a functional evaluation flight by sundown or to establish adequate limitations on ambient lighting and weather conditions led the flightcrew to attempt the stall series in the absence of a natural horizon. 17. There is a lack of consistency across the industry in the conditions and limitations for conducting functional evaluation flights and associated approach to stall maneuvers. 18. The informality of the ABX functional evaluation flight training program permitted the inappropriate pairing of two pilots for a functional evaluation flight, neither of whom had handled the flight controls during an actual stall in the DC-8. 19. The occurrence of fatal accidents during two different nonroutine operations (a functional evaluation flight and a three-engine ferry) by air carriers indicates a need to identify other nonroutine operations conducted by air carriers that may require additional procedural definition and training measures. 50 20. The flightcrew’s decision to conduct the flight at night was influenced by the succession of delays they had experienced earlier in the day. 21. The deficiencies of the ABX functional evaluation flight program remained latent after general organizational problems were identified by the 1991 National Aviation Safety Inspection Program in the other company functions. 22. The response of the FAA principal operations inspector to the 1991 ABX DC-8 loss-of-control incident was timely and appropriate, but it was not formally incorporated into ABX procedures. 23. The currently established FAA airworthiness and operating procedural requirements for conducting functional evaluation flights on large transport aircraft provide inadequate guidance to air carrier operators, maintenance repair stations, FAA principal operations and maintenance inspectors, and other affected parties. 51 3.2 Probable Cause The National Transportation Safety Board determines that the probable causes of this accident were the inappropriate control inputs applied by the flying pilot during a stall recovery attempt, the failure of the nonflying pilot-in-command to recognize, address, and correct these inappropriate control inputs, and the failure of ABX to establish a formal functional evaluation flight program that included adequate program guidelines, requirements and pilot training for performance of these flights. Contributing to the causes of the accident were the inoperative stick shaker stall warning system and the ABX DC-8 flight training simulator’s inadequate fidelity in reproducing the airplane’s stall characteristics. 52

[ANALYSIS] However, based on weather satellite data and almanac information, the Safety Board concludes that the flightcrew did not have a clearly visible natural horizon because of darkness and clouds above and below the airplane, and that the airplane most likely encountered IMC soon after descending through 13,500 feet and remained in IMC until just before impacting terrain. As a result of their initiating the maneuver with minimum clearance from clouds, the flightcrew was immediately forced to rely on instrument references for the stall recovery when the airplane entered a full stall and started a steep descent. The Safety Board recognizes that the artificial horizon on the electronic flight instrument system (EFIS) display facilitates precise attitude control and orientation in extreme pitch and roll attitudes. However, the natural horizon may have provided a more rapid orientation for the flightcrew in the range of +10/-20 degrees pitch and +/-70 degrees roll, which the airplane experienced during the early stages of the loss of control, because a natural horizon would have been visible through the airplane’s windshield in this range of pitch and roll attitudes. Other air carriers and manufacturers require that a natural horizon be visible during approach to stall or stall characteristics checks. The Safety Board concludes that by conducting these maneuvers without a visible natural horizon, the flightcrew was deprived of an important flight attitude reference that would have aided in their recovery from a full stall. The role of ABX in regard to this issue is discussed in section 2.5.2. 40 2.4 Simulator Fidelity and Training for Stall Recovery Procedures The Safety Board’s evaluation of the ABX DC-8 simulator indicated that the simulator did not reproduce the stall characteristics of the DC-8 with fidelity. For example, when slowed to below the airspeed of stick shaker activation, the simulator developed a stable, nosehigh, wings-level descent, with no tendency to pitch down in a stall break (abrupt nose-down pitch or roll). In contrast, according to Douglas and ABX manuals and the FDR data from the accident flight, the actual DC-8 airplane’s stall characteristics include a pronounced stall break. Further, after slowing well below stall speed, the simulator entered a mode in which the aerodynamic buffet stopped and the airspeed did not continue to decrease. The simulator’s benign flight characteristics when flown more into the stall provided the flightcrew with a misleading expectation of the handling characteristics of the actual airplane. The PF’s initial target pitch attitudes during the attempted stall recovery (from 10 degrees to 14 degrees) may have resulted in a successful recovery during his practice and teaching in the simulator. Further, because their experience with stalls in the DC-8 was obtained in a simulator without a stall break, the PF and PNF could not practice the nose-down control inputs required to recover a stalled airplane that is pitching down or at a nose-low attitude. Moreover, because the PF and PNF were exposed during extensive simulator experience to what they presumed was the stall behavior of the DC-8, the stall break that occurred in the airplane most likely surprised them.

[ANALYSIS] However, that log page was ambiguous as to who had been the PIC of the December 21 flight because the PNF signed for the airplane in the “Captain’s signature” block. The aircraft log page for the accident flight was not recovered. The PNF had previous experience as the SIC of several DC-8 post-modification FEFs, while the FEFs of December 21 and 22 were the first exposure of the PF to postmodification FEF procedures. Although it is typical for the left seat pilot to fulfill the role of the PIC, flight instruction can be conducted from the right seat, and the flight instructor usually serves as the PIC. Flightcrew comments recorded on the CVR during the set-up for the clean stall maneuver and the unpowered controls Vmc 35 check that preceded it, revealed that the PNF directed the PF, while the PF expressed uncertainty and asked questions. Further, several of the communications from the PNF to the PF were instructional in nature. Based on these communications and the PF’s lack of any prior experience with a complete DC-8 postmodification FEF, the Safety Board concludes that the PNF, in the right seat, was serving as the pilot-in-command of the accident flight and was conducting instruction in FEF procedures; and that the PF, in the left seat, was serving as the second-in-command and was receiving instruction in FEF maneuvers, including the clean stall maneuver. During the attempted stall recovery, there were several indications of the PF’s excessive aft control column inputs that should have suggested to the PNF that, as the PIC, he needed to correct the control inputs and recover from the stall. These included the position of the control column, which was, at times, being held in the full aft position by the PF; continued aerodynamic buffet; the extreme pitch down moments (stall breaks) accompanied by roll-off into steep bank attitudes; engine compressor surges; and the instrument indications of low airspeed and high rate of descent. The Safety Board evaluated why, despite these cues, the PNF did not take control of the airplane or otherwise intervene effectively as the PF held the airplane in a stalled condition all the way to impact. The reduced effectiveness of the monitoring and challenging of one crew member by another when each is qualified as captain has been recognized in previous accidents investigated by the Safety Board.36 In this accident, both pilots were captains, both were managers, and both had similar backgrounds at ABX. In this kind of crew pairing, it may be difficult for one captain to challenge the actions of the other because of a lack of overt command authority. The PNF’s role as an instructor pilot during the FEF maneuvers should have clarified the roles and responsibilities of the two pilots.

[ANALYSIS] The prospect of conducting the stall series at night should have at least prompted the flightcrew to consider whether additional risks were involved, both prior to departure and while preparing to perform the stall maneuver. The Safety Board evaluated whether the flightcrew’s decision to undertake the FEF at night was prompted by supervisory or self-imposed pressure. Completion of modifications to N827AX had been delayed for several months. Although ABX operational and maintenance managers said that the company had made no specific plans to use the airplane in revenue service, ABX marketing managers indicated that the delayed availability of the airplane had caused the company to inform a freight charter customer that its charters were subject to cancellation on short notice (because N827AX would not be available in case another airplane needed repairs) and subsequently to cancel one of these charters. In addition, the loss of the airplane in the accident caused ABX to lose another charter opportunity. The Safety Board was unable to identify the accident flightcrew’s state of awareness of these specific plans for N827AX; however, according to ABX’s director of flight training and standards, the accident PNF was aware of the company’s desire to place the airplane 45NTSB/AAR-95/06. op. cit. 46 in revenue service. The flightcrew, as ABX managers, would most likely have responded to the urgency to place the airplane in service with a strong effort to get the job done. However, there was no evidence of direct pressure on the flightcrew from higher level ABX managers to complete the flight. Further, on the day of the accident, the flightcrew had experienced a succession of delays as maintenance was completed on the airplane. The crew must have only gradually become aware, as the delays extended, that the planned evaluation flight could not be completed in daylight conditions. For the flight’s originally scheduled 1320 departure time and for several of the subsequent estimated departure times, the flightcrew would have assessed that they could complete the flight in daylight; after having made these initial assessments, it may have been more difficult for them to reverse their decision to perform the flight. The Safety Board concludes that the flightcrew’s decision to conduct the flight at night was influenced by the succession of delays they had experienced earlier in the day. 2.6 FAA Oversight 2.6.1 FAA Surveillance of ABX The FAA’s 1991 NASIP inspection found evidence that ABX had problems ensuring that company manuals and other documentation of operations procedures kept pace with the company’s growth. There were no requirements for the FAA to provide surveillance of the FEF program, and there was no documentation on the FEF program in the NASIP report. However, this accident has indicated that the ABX FEF program was also functioning with inadequately defined and documented guidelines for conducting evaluation flights, and inadequate flightcrew training and qualification standards for FEFs. Therefore, the FEF program deficiencies were consistent with the general problems identified by the 1991 FAA inspection. Subsequent action by ABX and the FAA to resolve these general deficiencies focused on the specific operational areas identified by the NASIP inspection.

[ANALYSIS] In this kind of crew pairing, it may be difficult for one captain to challenge the actions of the other because of a lack of overt command authority. The PNF’s role as an instructor pilot during the FEF maneuvers should have clarified the roles and responsibilities of the two pilots. As the instructor, his pilot-in-command authority should have been enhanced, and it should not have been difficult for him to exercise direct control to ensure the safety of the flight. However, the PNF’s instructional role, like his 35Vmc is defined as minimum controllable airspeed with critical engine inoperative. 36For example, see Aircraft Accident Report—“Midwest Express Airline, Inc., DC-9-14, N100ME, General Billy Mitchell Field, Milwaukee, Wisconsin, September 8, 1985” (NTSB/AAR-87/-1); and Aircraft Accident/Incident Summary Report—“Controlled Flight into Terrain, GP Express Airlines, Inc., N115GP, Beechcraft C99, Shelton, Nebraska, April 28, 1993” (NTSB/AAR-94/01/SUM). 36 command role, was informal on this flight. Therefore, his command and instructional authority may have remained unclear. Further, instructors often prefer to “talk” a student through a recovery, rather than take over the controls. This milder, verbal intervention can be carried too far if instructors are overconfident about their students’ or their own ability to recover at the last moment, or if they lose their awareness of impending danger. In this accident sequence, early in the recovery attempt at 1808:30, the PNF made a helpful comment to the PF, “You can take a little altitude down, take it down [that is, give up altitude to facilitate the stall recovery].” However, when the PF did not respond adequately, the PNF did not escalate his verbal interventions or take over the controls. During the uncontrolled descent, the PNF continued to provide suggested control inputs to the PF, but the PNF never told the PF to move the control column forward. It is possible that the PNF’s priorities changed after the extreme rolling moments developed. The PNF may have become less concerned about angle of attack, and the reduction of pitch attitude may have become a secondary priority to roll attitude control. Throughout the recovery sequence, the PNF made several statements to the PF to help direct him out of the roll situation (but not out of the stall condition), and he remained in an instructional role up to the time of impact. During this period, the PNF also took time to respond to a query from ATC pertaining to the airplane’s rapid descent below its assigned block altitude floor. Based on the PNF’s lack of urgency in correcting the PF’s control inputs and the PNF’s radio transmissions with ATC, he apparently lost awareness of the flight’s descent rate and proximity to the ground.

[ANALYSIS] The Safety Board was unable to determine the extent to which these possible conditions individually contributed to the early onset of buffet, but concludes that some combination of airframe icing, flight control rigging, or other factors resulted in the greater-than-expected buffet onset speed. However, the Safety Board’s analysis of FDR parameters indicated that the actual stall occurred at 126 knots, only four knots greater than the stall speed calculated by the flightcrew.33 Therefore, despite the early onset of aerodynamic buffet, the Safety Board concludes that any effects of airframe icing or flight control rigging upon the stall speed of the accident airplane were minimal, and did not contribute to the accident. 2.2.3 Flightcrew’s Mis-trim of the Airplane At 1807:43, when the FDR data indicated the airspeed was decreasing through 173 knots, the PF stated, “Guess I better not trim below * two *.”34 However, ABX procedures for the clean stall maneuver required that he stop trimming the airplane’s nose up at 1.5 Vs (184 knots for the accident flight). Based on the measured position of a horizontal stabilizer trim jackscrew found at the accident site, the stabilizer was trimmed to the 8-degree ANU position. Simulations and calculations conducted by Douglas based on the accident airplane’s weight and CG location show that an 8-degree ANU trim would result in a trim speed of about 175 knots. The effect on the stall recovery of trimming the airplane to a slower airspeed than desired would have been to require more forward control column movement to achieve the desired nose-down pitch rate and recover from the stall. However, because the airplane’s 175knot trim speed was significantly greater than the 126 knot stall speed, additional control column back pressure was required from the PF to slow the airplane to the stall speed, and the airplane would have rapidly recovered from the stall if the PF had relaxed his back pressure on the control column. Therefore, the Safety Board concludes that although the PF trimmed the airplane below the recommended minimum trim speed for the clean stall, this action did not contribute to the accident. 2.2.4 Monitoring and Challenging by the PNF Because both pilots were qualified to act as a DC-8 captain at ABX, the Safety Board evaluated flight records, the flightcrew’s previous experience with check flights and flightcrew statements on the CVR to determine whether the PNF was serving as the pilot-incommand of the accident flight and instructing the PF from the right seat. The ABX aircraft log page for the FEF of December 21, during which the two pilots had flown in the same seating positions as on the accident flight, had the PF’s name 33According to FDR data, an aerodynamic stall occurred between 18:08:21 and 18:08:22. 34When transcribing CVR recordings, the Safety Board uses an asterisk to signify an unintelligible word. 35 entered in the “Captain” block.

[ANALYSIS] During this period, the PNF also took time to respond to a query from ATC pertaining to the airplane’s rapid descent below its assigned block altitude floor. Based on the PNF’s lack of urgency in correcting the PF’s control inputs and the PNF’s radio transmissions with ATC, he apparently lost awareness of the flight’s descent rate and proximity to the ground. The Safety Board concludes that the PNF’s failure, as the pilot-in-command, to recognize, address and correct the PF’s inappropriate control inputs were causal to this accident. 2.3 Cues Presented to the Flightcrew The Safety Board evaluated the adequacy of the cues that were presented to the flightcrew and the role of distraction as possible contributors to the flightcrew’s failure to recover from the stall. The Board examined the cues that were presented to the flightcrew by the airplane’s flight deck instrumentation and engines, and the flightcrew’s visual references to the natural horizon based on the existing weather and light conditions. 2.3.1 Stall Warning System The stall warning system stick shaker failed to activate during the accident sequence at the appropriate margin above the stall and during the full stall that followed. According to Douglas, DC-8 models up to the -61 series provided an adequate margin between the buffet and stall in the clean configuration, so they did not require the artificial stall warning of the stick shaker. However, the stick shaker was needed to fulfill the certification requirements for adequate pre-stall warning on subsequent DC-8 models, including the accident airplane (a DC-8-63). 37 Despite these requirements, ABX FEF pilots told Safety Board investigators that the stick shaker activated inconsistently relative to the onset of pre-stall buffet during the clean stall maneuver, sometimes activating prior to buffet, but in some cases activating simultaneously with buffet or after the buffet had begun. According to the ABX director of flight technical programs, FEF flightcrews were told that they should be prepared for the stick shaker not to activate before the stall during the clean stall maneuver. Further, the CVR transcript indicates that the flightcrew had clearly identified a stall buffet and that they initiated a recovery in a timely manner. Consequently, the Safety Board concludes that the absence of the stick shaker prior to the stall did not affect the flightcrew’s recognition of the initial entry into the stall. However, the Safety Board is concerned that the stick shaker failed to activate at any time during the full stall that followed. Given the extreme angles of attack that were eventually attained prior to impact, the stall warning system clearly was inoperative. The Safety Board attempted to determine why the airplane’s stick shaker stall warning system did not activate. Douglas records indicated that the airplane’s stall warning system functioned in accordance with its design specifications when it was delivered in 1967. Preflight cockpit checks before the accident flight indicated that the airplane’s stall warning system was operational; however, the preflight check of this system (using the test switch located on the flight deck overhead panel) does not test the stall warning system’s wing-mounted angle of attack sensor and transducer.

[ANALYSIS] According to ABX records, the DC-8 flight standards manager performed most of the DC-8 postmodification FEFs (which involved stalls) from 1991 through 1994, while the director of flight technical programs performed a limited number during that period. The POI recalled that at some time following the 1991 incident (he did not remember precisely when) the director of flight technical programs had voiced disagreement about the need to change the FEF stall recovery procedures. The POI said that he was not 42ABX records indicate that the director of flight technical programs was on extended sick leave for some periods in 1989 and 1990, and he did not log flight time at ABX between February 19, 1990, and February 14, 1991. From March 1991 to August 1992, he was extensively involved in the ABX DC-9 FEF program. He returned to DC-8 flying in August 1992, and conducted two post-modification FEFs in the DC-8 through 1994. 42 concerned about this disagreement because, at that time, the director of flight technical programs had a limited role in the DC-8 FEF program. The director of flight technical programs also confirmed to Safety Board investigators his belief that the stall recovery procedures contained in ABX’s operations manual were adequate for FEFs if performed with sufficient attention to engine spooling. He had continued to use these stall recovery procedures when he flew FEFs. After the DC-8 flight standards manager, who had adopted the revised FEF procedure, returned to line flying in 1994 and was replaced by the accident PNF, the director of flight technical programs trained the accident PNF using the original stall recovery procedure. The accident PNF, in turn, trained the accident PF. The Safety Board notes that some provisions of the revised procedure were implemented and used during the accident flight, although the manner in which they were used confirmed that the flightcrew was attempting the original, minimum altitude loss stall recovery. For example, the flightcrew used the block altitude clearance designed to provide sufficient altitude for stall recoveries involving greater altitude loss. However, they only obtained a 2,000foot block, rather than the 3,000-foot to 5,000-foot block specified in the revised procedure. In addition, they conducted the stall 500 feet above the bottom of the altitude block and only slightly above the clouds. The flightcrew also referred to engine spooling. Further, although both pilots received simulator training during their informal qualification as FEF pilots, ABX did not implement a simulator session prior to each FEF, which was also specified by the POI. Despite its partial implementation of the revised procedure, the statements of the director of flight technical programs and the actions of the accident flightcrew show that ABX ultimately did not institutionalize the technique of exchanging altitude for a more rapid stall recovery, and thus failed to take advantage of the valid lessons of the 1991 DC-8 FEF loss-ofcontrol incident.

[ANALYSIS] The data showed that liftoff occurred at a slightly higher-thannormal airspeed and that the airplane began to climb. However, when reaching an altitude of 100 feet or less, the airplane rolled steeply. Although the tower controller reported seeing the airplane roll to the right and strike the ground in a nearly inverted attitude, the majority of witnesses and the physical evidence support a finding that the airplane’s left wing struck the ground first. 44 Four previous accidents have been investigated by the Safety Board in which it was determined that DC-9 Series 10 airplanes have encountered nearly identical flight control difficulties during takeoff in conditions conducive to the accumulation of wing airfoil ice contamination. The investigation of this accident provides substantial evidence that the rapid roll and descent after Jiftoff were the result of an aerodynamic stall. As in the previous accidents, the airplane was able to Vift off and climb initially because of the influence of ground effect on the aerodynamic characteristics of the wing. When an airplane is close to the ground plane, the direction of airflow over the wing is altered. The result is that the wing will produce more Vift at the same airspeed and AOA than it will when the airplane is in free air. This enhanced aerodynamic performance diminishes as the airplane climbs and becomes almost negligible at a height equal to the airplane’s wingspan, a distance of 87 feet for the DC-9-15. Generally, an airplane’s rotation speed is selected so that, with a normal rate of rotation, the airpiane will lift off at a speed that offers a Safe margin above the stall speed. In this case, the first officer rotated the airplane at the proper airspeed (132 knots) to ensure this stall margin. Normally, the airplane would have become airborne 2 or 3 seconds later at an airspeed of about 142 knots, providing more than 20 knots of stall speed margin. However, the FDR data showed that Ryan 590 lifted off about 4 seconds after the "Rotate" call and, assuming that the first officer rotated the airplane smoothly at around 39 per second in accordance with the Ryan Operators Manual, the airplane was probably at a higher-than-normal pitch attitude with a correspondingly higher-than-normal AOA of 10° or greater before it became airborne. Because there was a faint scar on the runway surface about 3,440 feet from the beginning of the runway and an indication that the tail skid of the airplane had at some time been in contact with a hard surface, the Safety Board considered the possibility of a tail strike occurring during the takeoff, thereby placing the airplane at an AOA of even more than 10 before or at liftoff. According to the manufacturer, to strike the tail with the wheels on the ground, the airplane would have to be rotated to a nose-up attitude between 11.5° and 15.5° depending upon the extension of the main gear struts.

[ANALYSIS] The Safety Board also believes that factors which contributed to a lack of flightcrew guidance on the importance of such inspections and the flight characteristics of the DC-9 Series 10 airplane, in particular, were causal factors in the accident. 48 The Safety Board believes that after failing to detect and remove the accumulation of ice from the wing, there were no actions that the crew could have been reasonably expected to take that would have prevented this accident. The first officer followed the normal and prescribed procedures for the takeoff; that is, the rotation speed was that specified for the airplane’s weight and the rotation rate was normal. When the airplane became airborne with a minimum stall speed margin, the stall was inevitable as the aerodynamic advantage of ground effect diminished. further, the stall was most likely more sudden and severe than would have occured with an uncontaminated wing because a stall can progress from the wing tips inward. This causes the airplane to pitch nose up with a loss of rol? control. The abrupt roll, occurring as one wing stalled before the other, was not controllable within the altitude available. Under the circumstances for the takeoff of Ryan 590, it might have been possible to increase the liftoff speed stall margin and establish a climb without stalling by delaying the takeoff rotation, permitting additional acceleration on the runway. However, this procedure would have been improper because the increase in the rotation speed beyond that specified may have infringed upon the safety margin required by the Federal Aviation Regulations (FARs) in case of an engine failure during the takeoff. The rotation speed is currently based upon a minimum field length takeoff for the airplane’s weight; that is, a field length that is sufficient to satisfy the balanced field concept where the accelerate-stop and accelerate-go distances are equal, assuming that an engine failure occurs at the decision speed, and also sufficient to satisfy the posttakeoff climb gradient requirement for obstacle clearance, as specified in the FARs. However, when operating on a runway longer than needed to meet this balanced or minimum field length criteria, a rotation speed higher than that currently specified could be used safely if the flightcrew were given sufficient information in their operating manuals to determine the maximum rotation speed that will still allow the required engine failure safety margins. The Safety Board believes that the FAA should require that this information be included in the manual to provide an additional takeoff safety margin for the DC-9 series 10 airplanes when they are operated from "unbalanced" runways in weather conducive to the formation of wing ice contamination, regardless of the other nece:sary measures to ensure that the wing is free of such contamination. 2.5 Dissemination of Airframe Icing Information The written material, industry presentations, and operator seminars that were offered for more than 20 years should have eliminated any operational problem with icing on the OC-9. However, similar accidents continue to occur. The Safety Board therefore concludes that efforts to educate line pilots of OC-9 series 10 airplanes about this problem have not been adequate. There are many reasons for the inadequacy of these efforts.

[CONCLUSIONS] CONCLUSIONS Findings RECOMMENDATIONS APPENDIXES Appendix A--Investigation and Hearing Appendix B--Cockpit Voice Recorder Transcript Appendix C--Personnel Information Appendix D--Additional Aircraft Information Appendix E--Douglas Aircraft Company Letter Report Appendix F--Douglas Aircraft Company Graphic J}}lustration . Appendix G--Company History Provided by Ryan International Airlines, Inc. EXECUTIVE SUMMARY About 0019, Sunday, February 17, 1991, Ryan International Airlines flight 590 (Ryan 590), a OC-9 series 10 airplane, crashed while taking off from Cleveland-Hopkins International Airport. The flightcrew consisted of two pilots. There were no other crewmembers or passengers on the flight, which was contracted to carry mail for the U.S. Postal Service. Both pilots were fatally injured, and the airplane was destroyed as a result of the accident. The Nationat Transportation Safety Board determines that the probable cause of this accident was the failure of the flightcrew to detect and remove ice contamination on the airplane’s wings, which was largely a result of a lack of appropriate response by the Federal Aviation Administration, Douglas Aircraft Company, and Ryan International Airlines to the known critical effect that a minute amount af contamination has on the stall characteristics of the DOC-9 series 10 airplane. The ice contamination led to wing stall and loss of control during the attempted takeoff. The safety issues discussed in this report include’ the dissemination of information regarding precautions to be taken when operating in conditions conducive to airframe ice and the particular susceptibility of DC-9 series 10 airplanes to control problems during take off when a minute amount of ice is on the wing. z t s £ i ¥ rf NATIONAL TRANSPORTATION SAFETY BOARD WASHINGTON, D.C. 20594 AIRCRAFT ACCIDENT REPORT RYAN INTERNATIONAL AIRLING DC-9-15, N565PC LOSS OF CONTROL ON TAKEOFF CLEVELAND-HOPKINS INTERNATIONAL AIRPORT, OHIO FEBRUARY 17, 1991 1. FACTUAL INFORMATION 1.1 History of the Flight About 0019, Sunday, February 17, 1991, Ryan International Airlines flioht 590 (Ryan 590), a ODC-9-15, crashed while taking off from Cleveland-Hopkins International Airport (CLE). The flightcrew consisted of two pilots. There were no other crewmembers or passengers on the flight, which was contracted to carry mail for the U.S. Postal Service. Both pilots were fatally injured, and the airplane was destreyed as a result of the accident.

[FINDINGS] The steep roll concurrent with stall] was caused by the irregular lift distribution across the wing and was not controllable by the pilot, thereby preventing recovery. The DC-9 series 10 has no Wing leading edge lift augment ing devices and is particularly vulnerable to degraded aerodynamic performance as 2 result of minute amounts of wing contamination than the later model DC-9 and MO-80 airplanes that have leading edge devices. The flightcrew had not been given specific training or other educational materia! to inform them of the more critical effects of wing contamination on DC-9 series 10 airplanes. The Doug’as Aircraft Company has issued numerous articles on the subject of wing contamination, but there is no system to ensure that the critical information reaches all line pilots of these airplanes. Both the FAA and Douglas Aircraft Company have been aware for several yeas of the propensity of the OC-9 series 10 to the loss of control caused by wing contamination, but neither of them took positive action to include related information in the approved Airplane Flight Manual. Had additional information or cautions about the high vulnerabil:ty of the DC-9 series 10 to loss of control caused by wing contamination been placed in the approved Airplane Flight Manual, it would have been available to line pilots. Ryan International Airlines had the Opportunity and obligation to request information relating to previously identified Safety issues when it acquired the OC-9 airplanes in 1989 but failed to do so. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the failure of the flightcrew to detect and remove ice contamination on the airplane’s wings, which was largely a result of a lack of appropriate response by the Federal Aviation Administration, Douglas Aircraft Company, and Ryan International Airlines to 52 the known critical effect that a minute amount of contamination has on the Sta'l characteristics of the DC-9 series 10 airplane. The ice contamination led to wing stall and loss of control during the attempted takeoff.

[FINDINGS] If Ryan had fulfilted this obligation it would have become aware of the previous accidents involving wing ice contamination. Then Ryan would have been able to provide the training and guidance to its flightcrews that should have prevented this accident. Thus, the airline is also cited as a causal factor in the accident. 50 3.0 CONCLUSIONS 3.1 Findings J}}. There was no evidence of preexisting airplane structural, systems, or engine faults that contributed to the loss of contro) of the airplane 7 seconds after liftoff from runway 23L at Cleveland-Hopkins International Airport. 2. Four previous accidents of DC-9 series 10 airplanes, also invelving loss of control almost immediately after liftoff, were attributed to a loss of aerodynamic efficiency due to ice accumulation on the wings. 3. The accident airplane had flown through conditions conducive to the accumulation of moderate rime ice during the descent for landing at Cleveland-Hopkins International Airport, about 40 minutes before the accident, and the flightcrew probably used the wing anti-ice system during the descent. 4. Ground conditions at the Cleveland-Hopkins International Airport during the 35-minute turnaround were not conducive to airframe icing because of the dry snow and the low ambient temperatures; however, the melting and refreezing of snow on the previously heated wings could have produced = an accumulation of ice on the wing upper surface. The flightcrew did not exit the airplane to conduct an exterior preflight inspection at the Cleveland-Hopkins International Airport to verify that the wings were free of ice contamination, and a requirement for such an inspection was not specified in the Ryan DOC-9 Operations Manual. The first officer was controlling the airplane, and the takeoff rol] and rotation were normal and accomplished in accordance with prescribed procedures. 7. The liftoff occurred at a _ higher-than-normal airspeed; however, the lift-producing efficiency of the wing was degraded by contamination, and the stall speed margin at liftoff was minimal. There was some physical evidence but no evidence derived from the performance analysis to corroborate a tail strike at takeoff. However, a tail strike could occur with normal pilot procedures during an attempted takeoff with wing contamination. 51 The airplane’s wings stalled abruptly and without warning as the airplane began to climb and the aerodynamic advantage of ground effect diminished. At the time of stall, the airplane had sufficient speed to achieve a 1.4 g load factor with normal aerodynamic characteristics. The engine compressor Surges were caused by the disturbed airflow aft of the wince and at the engine inlet as the airplane approached stall. The steep roll concurrent with stall] was caused by the irregular lift distribution across the wing and was not controllable by the pilot, thereby preventing recovery. The DC-9 series 10 has no Wing leading edge lift augment ing devices and is particularly vulnerable to degraded aerodynamic performance as 2 result of minute amounts of wing contamination than the later model DC-9 and MO-80 airplanes that have leading edge devices.

[ANALYSIS] Although this instructor-student form of interaction did not warrant their deviation from company standard procedures, it nevertheless could have contributed to it, and provides a plausible explanation for some of the deviations. In summary, the Safety Board acknowledges that conditions were conducive to producing fatigue in the pilots and increasing susceptibility to disorientation. The accident circumstances certainly reflect substandard human performance; however, the available evidence is not sufficient to warrant a conclusion that crew fatigue adversely affected pilot performance in this accident, although the Safety Board cannot rule out that possibility. Additionally, there are other factors that could have contributed to the loss of control that either outweigh or complement fatigue as a factor in the accident. 2.9 Loss of Control According to the FDR, the captain began a slow, sustained left tum (about 0.6 degrees of heading per second and about 5 degrees of bank) about 0324:50. This action was contrary to the published missed approach instructions but possibly in anticipation of a downwind turn instruction identical to the one given by the controller on the previous missed approach. Beginning about 0325:00, the sound of a power reduction is ford on the CVR, and the airplane was approaching its 52 assigned altitude. However, at 0325:10, the FDR still showed the airplane ascending through 2,800 feet at a rate of about 2,400 feet per minute. At 0325:31, the FDR showed that the airplane's altitude peaked out at 211 feet above the assigned altitude of 3,000 feet. It is probable that the captain then realized he had overshot his assigned altitude and proceeded to push the nose over during the decelerating tum to regain 3,000 feet. About 5 seconds later, shortly after the first officer acknowledged the tum to 300 degrees, the FDR shows that the tum rate increased dramatically. Simulations show that the bank angle then steepened to about 25 degrees when the captain uttered the “What's the matter" comments. The flightpath study indicates that 8 seconds after exceeding 30 degrees bank angle, the airplane was passing through about 60 degrees left bank at a 14 degree descent angle. This obvious loss of control, in context with the CVR comments, prompted the Safety Board to examine the role played in the loss of control by: 1) transfer of control; 2) spatial disorientation; and 3) the attitude references. 2.9.1 Transfer of Control There was a positive transfer of control as the second ILS approach was aborted, when the captain stated, "Oh # I got it." While the first officer never acknowledged the release verbally, there were positive affirmations in his response to the captain's commands for flaps and gear. The first officer also began all radio calls as the nonflying pilot. However, there was no subsequent positive verbal transfer of control to the first officer prior to the loss of contro) during the second go-around.

[ANALYSIS] The first officer also began all radio calls as the nonflying pilot. However, there was no subsequent positive verbal transfer of control to the first officer prior to the loss of contro) during the second go-around. Even if the captain thought he had relinquished control to the first officer pricr to the control loss, the first officer demonstrated that he considered himtci: the nonflying pilot when he acknowledged the assigned vector of 300 degrees at 13°.:36.3. This transmission was made 2.6 seconds before the captain's first question, “What's the matter." In the subsequent 10 seconds until the captain asked, "You got it?” the situation deteriorated drastically. Although it is possible that neither pilot was flying the airplane during this period, given his almost constant prompting on the previous ILS approaches, it is ir probable that the captain allowed the first officer to fly so poorly without interjecting some instruction or taking control of the airplane again. Also, the positive initiation of the bank angle was not consistent with uncommanded airplane motions, which suggests that someone was on the conirols. 2.9.2 Spatial Disorientation The first officer's performance on the second ILS resulted in the captain taking control probably after not having controlled the airplane since the last landing in Portland. Orcgon, the previous day. There is no question that the captain became gravely concemed about something 1 minute and 22 seconds after assuming airplane control, almost immediately after receiving ATC instructions to tum to a heading of 300 degrees, about 22 seconds prior to impact. He asked, "# # what's the matter?” 2.5 seconds after the first officer routinely acknowledged the new heading assignment. About 4.5 seconds later, the captain became very alarmed and essentially repeated his initial remark. While both comments are nonspecific, the Safety Board considered the possibility that they reflected a state of perplexity and confusion, rather than the captain's analysis of a mechanical problem with the airplane. Although the captain was routinely turning to the assigned heading of 300 degrees when control was lost, he had just transitioned from a climb, reduced power and was still attempting to level at the assigned altitude of 3,000 feet. This combination of steady, sustained tuming, acceleration-to-deceleration changeover, and abrupt ascent to descent transition, at night with no visible horizon or outside references, is especially conducive to spatial disorientation. Medical experts state that deceleration while tuming can produce the sensation of tuming in the opposite direction. Had the captain of flight 805 incorrectly believed, at about 0325:36, that he was rolling out of his tum to the left (the opposite direction), he might have increased the bank angle to the left to compensate. The fact that the airplane was not simultaneously decelerating, turning, and descending for a very long time before the problem occurred could lessen the possibility of this phenomenon.

[ANALYSIS] Because of these discrepancies and the current state-of-the-art interpretation of instrument impact markings, the Safety Board was unable to determine which pitch witness marks occurred at initial ground impact and which marks were made during secondary impact(s). Therefore, the Safety Board cannot positively determine whether the captain's ADI was malfunctioning and was a causal element in this accident sequence. Nor can the Safety Board rule out the possibility that the ADI malfunctioned and precipitated the loss of control by the captain. 2.10 The Recovery Attempt There is little doubt that there was accurate attitude information available to the crew, at least from the standby artificial horizon. The display half of the standby artificial horizon displayed an attitude corresponding to that of impact. Further, it is believed that the first officer's ADI was accurate because of his recovery efforts. His response to the captain's release of control was not only immediate (0.7 36 seconds), but it was correct in general execution. The flightpath study indicates that at the time control of the airplane was transferred, the bank angle had increased to about 65 degrees, and the flightpath angle was about -15 degrees. The altitude at the time was about 3,100 feet. The Safety Board's research indicates that the descent angle continued to steepen. Within 4 seconds of the first officer stating, I got it,” the angle of the bank stabilized at about 80 degrees, the altitude was about 2,600 feet and the flightpath angle was about -28 degrees. The first officer correctly concentrated on trying to level the wings for the next 3 seconds. Then, as the bank angle approached 20 degrees, he began focusing on arresting the descent. Eventually, in the 12 seconds after he acknowledged assuming control, he recovered the airplane to the 15 degrees left bank, and the -17 degrees flightpath present at the time of impact. There are several points to be emphasized in evaluating the recovery attempt. Airline pilots are not periodically trained to recover from unusual attitudes as are military pilots or civilian acrobatic pilots. The presumption is that an airline pilot should avoid an unusual attitude and will never have a need to recover from one. Similarly, transport-category aircraft operations contrast with the military and civilian acrobatic method of flying an airplane. Airline pilots may subconsciously avoid abrupt maneuvers in the interest of passenger comfort. In this regard, the first officer had no military and little, if any, acrobatic flight training. Although the captain was a retired Naval Aviator, he last flew a military airplane in 1979, about 13 years prior to the accident. There was no indication recorded on the CVR that the captain was physically involved in the recovery maneuver at all. Once he relinquished control, he did not even "coach" anymore, until 2 seconds before impact when he joined the flight engineer in saying, “Up, up...." The first officer took control of the airplane at approximately 0325:50.

[ANALYSIS] There was no indication recorded on the CVR that the captain was physically involved in the recovery maneuver at all. Once he relinquished control, he did not even "coach" anymore, until 2 seconds before impact when he joined the flight engineer in saying, “Up, up...." The first officer took control of the airplane at approximately 0325:50. However, the captain's difficulties were signaled about 10 seconds before he asked “You got it?" The Safety Board believes that the captain verbalized his confusion at 0325:39 with his "What's the matter?” query. Also, the airplane exceeded a 30degree bank angle at around 0325:41. Because 30 degrees is the steepest bank angle used in normal! transport fiying, the captain's continued roll into a steeper bank should have alerted the first officer that he needed to challenge the captain's performance. The Safety Board believes that the accident might have been prevented if the first officer had corrected or challenged the captain's overbank in the 10 seconds between the first signal of trouble and the captain's transfer of control statement. The Safety Board was unable to determine the moment when the first officer detected that the captain was overbanking and unable to control the airplane. 57 Likewise, the Safety Board was unable to determine whether the first officer's poor performance on the preceding ILS approaches made him hesitant either to speak up and alert the captain to his deviation from normal flight attitude or to intervene and correct it. However, the Safety Board believes that this accident highlights the need for active crew coordination and interaction to avoid having the flying pilot exceed flight limitations, such as airspeed, pitch, and bank angles. The circumstances further emphasize the importance of timely action in challenging or correcting fellow crewmembers. Lastly, the basic control manipulations by the first officer during the recovery attempt were in general accordance with accepted procedures in that he attempted to roll the wings level and then began pulling the nose up. If he had been more aggressive with both sets of controls, he might have succeeded. A larger, more rapid aileron input would have leveled the wings faster; and a more aggressive pullout could have been within the operating envelope of the aircraft. Even if he had exceeded the approved g load for the DC-8, a large safety margin existed to preclude structural failure in extreme situations. Obviously, this situation called for extremely quick and aggressive control inputs. 2.11 Cockpit Resource Management on Flight 805 The Safety Board notes that the manner in which the flightcrew interacted in the latter stages of this flight was not consistent with .idely accepted CRM principles. Although it was impossible to determine how the pilots related to each other during their previous flight segments, activities in the cockpit for the last 30 minutes of the flight were more representative of an instructor/student than a teamwork situation. Cockpit conversation and interaction were one-sided, in that the captain was dominating the conversation and making all the decisions conceming the flight until the first officer assumed control of the airplane after the loss of control.

[ANALYSIS] AJT701 did not experience any problem on the second approach. ATI 815 landed at 0117, and the first officer noticed the captain's flight director oscillate from one dot fly up to one dot fly down. It did not affect the approach, and they made a normal landing. The ILS anomalies experienced by the crews appear to be so random that they could be explained by many variations, both in the transmitter and the receiver/displays. Based on tre information presented, some of the needle movement might be the normal fluctuation when the receiver first responds (particularly in ATI819 and ATI815). The two most perplexing problems were those of both B-727100s (AJT 803 and AJT 701). The problem that AJT701 had with the ILS was sufficient to cause that flight to abort the approach. Despite extensive effort, the Safety Board could not determine the cause for the problems noted with the ILS. There was no discussion of glideslope indication problems on the CVR during the approaches, and the crew reported that they, "...lost the localizer course...we had the glidepath but not the localizcr." In addition, ground and airborne checks of the ILS equipment by the FAA following the accident revealed no anomalies. Accordingly, the ILS is not considered to have been a factor in the accident. 59 3. CONCLUSIONS Findings 1. The flightcrew was properly certificated and qualified for the flight in accordance with existing regulations. 2. The airplane was properly certificated and maintained in accordance with existing regulations. 3. The first officer was flying the airplane upon arrival into the Toledo area. For undetermined reasons, he failed to properly capture the ILS localizer and/or glideslope during two ILS approaches. 4, The captain assumed control of the airplane during the second missed approach; however, he apparently became spatially disoriented, from physiological factors and/or a failed attitude director indicator, and he inadvertently allowed an unusual attitude to develop with bank angles up to 80 degrees and pitch angles up to 25 degrees. 5. The captain transferred control of the airplane to the first officer when the airplane was nose low and in a left bank angie; however, there may have been a short period of time when neither pilot was in control. 6, The first officer assumed control and began leveling the wings and raising the nose of the airplane, but impact with the ground occurred before the unusual attitude recovery was completed. 7. The operability of the captain's attitude director indicator at the time control was lost is uncertain. Witness marks on the one attitude director indicator ball that was found could have indicated an incorrect position at impact; however, the evidence was inconclusive. 8. Based on the observed performance of the airplane during the recovery, the first officer's attitude director indicator was operating properly. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the failure of the flightcrew to properly recognize or recover in a timely manner from the unusual aircraft attitude that resulted from the captain's apparent spatial disorientation, resulting from physiological factors and/or a failed attitude director indicator.

[ANALYSIS] There were no crewmember comments on rudder feel or rudder direction. A hard-over rudder or a rudder jammed at an extreme angle would have produced a sudden high, constant heading change rate that would have been recorded on the FDR. The FDR, however, recorded varying heading change rates that were momentarily high but, overall, were consistent and smooth until impact. Thus, it is concluded that the loss of control did not result from any cargo shift or flight contro! problem. 2.3 The Asymmetric Flap Scenario The CVR indicated that the last flap callout was by the captain for 25 degrees. However, since the airplane had a 23-degree flap detent, rather than 25 46 degrees, and since the go-around flap setting was 23 degrees, the captain's intended flap setting was probably 23 degrees. Since the flap actuators do not necessarily retain their positions after hydraulic pressure is removed, the actuator stroke measurements in the wreckage cannot be considered a definitive indication of exact flap extension at impact. Therefore, the impact flap setting could not be positively determined. Asymmetrical flaps could cause the roll initiating the loss of control. However, the captain asked for the go-arcund flap setting at 0324:30, and the sound of flap handle movement is heard on the CVR seconds later. The first indication of trouble, the captain's question, "What's the matter?" was recorded I minute and 9 seconds after the flap handle sound. According to the manufacturer, the time to retract the flaps from full down (the landing setting) to 23 degrees of flaps is between 6 and 10 seconds. Any asymmetric flap problem should have been evident to the captain within seconds, and prompted a comment about the flaps. Accordingly, the Safety Board concludes that the departure from level “light was not initiate4 by a flap system problem. 2.4 The First Officer Seat Failure Scenario The crew that flew the airplane to POX entered the following writeup in the maintenance log: "F/O seat difficult to lock in fore and aft position.” The Safety Board considered this writeup to be significant because other accidents have resulted from a crewmember's unintentional manipulation of the flight controls due to an unexpected seat movement. The discrepancy was written off with corrective action. Although the first officer's seat was the subject of various complaints previously, there was no reference to seat problems on the CVR. In addiiion, the difficulty in flying the airplane began at a time when the captain was at the controls. The Safety Board therefore concludes that first officer's seat anomalies were not involved in the cause of the accident. 2.5 The Asymmetrical Engine Thrust Scenario The sound spectrum analysis of the CVR recording revealed an 8.2 percent of maximum thrust differential between two of the four engines on the airplane about 5 seconds before the captain's question "What's the matter?" Inherent limitations of a sound spectrum analysis precluded identification of the engine with the lowest and the engine with the highest amount cf thrust.

[ANALYSIS] Thus, the engine instruments should have confirmed the engine failure, and the flight instruments should have confirmed the airplane's attitude, airspeed, altitude, and heading. Nevertheless, the questions asked by the captain and his failure to maintain eontrol of the airplane confirm that he did not correctly interpret the sounds, notion, and other availab.e information. Therefore, the Safety Board believes that the captain reacted primarily to other than visual and flight instrument references, such as Xinesthetie cues. He apparently misinterpreted those cues and applied the flight controls incorrectly. This confusion is commoniy referred to as “spatial disorientation,” and it oceurs most frequently at night or in instrument meteorologica! conditions when few, if any, exernal visual references exist. Spatial disorientation causes confusion, such as that experienced by the pilots of flight 105, and would account for their incorrect control resconses. The only means to drevent such confusion, or to overcome its effects, is for the pilot to relv on the flight instruments. erefore, the Safety Board concludes that infrequent training for an engine failure at ton altitude in the initial climb phase of flight could bave left the flightcrew ili-repared to cope with the emergeney- Although analyzing abnormal or emergency Situations end maintaining control of the airplane by reference to flight instruments are basic elements of airmanship, the Safety Board believes that the FAA and the airline industry should consider the circumstances of this accident with a view toward including scenarios of engine failures after establishment of the takeoff climb in training programs to better prepare pilots for such emergencies. Consideration also should be given to reducing pilot. reHance on external visual cues Guring “V1 cut” training by making greater use of simulated low visibility situations during such training. 25.4 Crew Coordination Tre CVR comments suggest that the captain was uncertain and perheps confused by the events which immediately followed the failure of the right engine. He had never experienced an in-flight engine failure on a DC-~-S. and he had not heard the souncs associated with such a failure in his flight simulator training. The vaw and deceleration metion cues he felt in the airplane also would have been slichtiv different from the ones to which he had been exposed in his simulstor training. His first question to the first officer ("What the # was that?"), may have been rhetorical; however, the Board betieves that the captain was requesting assistance. His second question ("What do we got here, Bili"?), oceurred 3 seconds after the right engine failure and affirms the concern and uncertainty expressed in his first question.

[ANALYSIS] Returning the rudder to neutral and holding neutral rudder, after initially applying rudder to correct for differential engine thrust. wouid not have created the heading change rates which were indicated by the FDR data. Likewise, a system malfunction which would cause the rudder to trail in a near-neutral position would be inconsistent with the FDR-indicated heading change data. The demonstration flight in a DC-9-14 airplane showed that the airplane had no control characteristics which were inconsistent with the applicable certification standards; the airplane was feund to be fuy controliable in an engine-out flight environment, even without using rudder (the primary control for correeting vaw and maintaining heading) to correct for yaw. Having found no evidence or airplane performance basis for concluding that there was a control system failure or malfunction, the Safety Board concludes that the rudder deflection, which occurred beginning 4 to 5 seconds after the right engine failure, was the result of the flightcrew's improper response. Based on the analysis of the airplane performance, the yaw generated by the incorrect rudder deflection, combined with G loading, caused the airplane to enter an accelerated stall at an altitude tco low for recovery. In the seconds which preceded the accelerated stall and loss of eontrol, the airpiane was ina very dynamic situation. The increasing rate of roll, the sideslip, and the increase in acceleration load ali affected adversely the stall speed. Because of the rapidly changing attitude of the airplane, the pilots would not have been expected to know the speed at which the airplane would stall in aecelerated flight. Compared to the increase in stall speed, the 8-Knot error in indicated airspeed (due to static source error in a sideslip) would not have been significant. Further, the stickshaker stall warning system would not and did not provide the customary 4 to 5 seconds warning which is typical of that system because of the rapid entry into the stall. The Safety Board concludes that the stall occurred because the flighterew dic not diagnose the nature of the emergency correctly, applied ineorreet rudder control about 4 to 5 seconds after the right engine failure, and applied nose-up elevator control which increased the G loads. The nose-up elevator contro}} input would have been a normal response to correct for the pitech-over maneuver and the reduction in pitch attitude which was precipitated Dy the rudder pedal induced roll and was consistent with the rapid deceleration of the airplane. The rapid deceleration would have resulted in a vestibular perception of downward pitching of the nose of the airplane. The Safety Board believes that more effective scanning of the flight and engine instruments by the pilots of fight 105 would have enabled them to maintain control of the airolane and to properly evaluate the powerplant anomalies.

[ANALYSIS] The crack had propagated to a length which should have sllowed detection on the occasion of the last high pressure compressor overhaul and spacer rework in 1981. 8. None of the airplane flight control systems were disabled. 9. The cause cf the left engine power loss, which occurred beginning about 1.5 seconds after the right engine failed, was not determined. 19. The left engine experienced a compressor stall in the last seconds of the flight after contre] had been lost and the airplane was descending toward the ground in an unusual attitude. 11. The loss of teft engine power was not significant with respect to the loss of control of the airplane. 12. 13. 14, 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. ~68- The captain initially responded correctly with deflection of the rudder pedal to the left to compensate for the loss of right engine thrust and by lowering the nose of the aircraft; however, he appeared to be unaware of the exact nature of the emergency. The crew response to the right engine failure was not coordinated. Neither pilot verbally identified the emergency condition or made the emergency callouts required by FAA-approved Midwest Express procedures. The rudder was incorrectly deflected to the right 4 to 5 seconds after the right engine failure. An accelerated stall and loss of control occurred 10 seconds after the failure of the right engine. Forward visual cues (outside the cockpit) were nut available to the crew at the time that the right engine failed. Peripheral visual cues were available. The visual flight simulator, which wes used by the crewmembers in training, did not provide onset y-w aad longitudinal acceleration cues, peripheral visual cues, or aural cues which were available to the crew in the airplane. The captain and first officer misinterpreted the inside visual cues which were presented in the airplane. The differences in visual motion and aural cues presented in the visual flight simulator and in the airplane may have limited the ability of the flighterew to recognize and react ar-oropriately to the emergency. Failure to recognize the nature of the emergency and improper operation of flight controls precipitated the loss of control. The DC-9-14 does not require unusual pilot skill or strength to maintain continued flight foliowing an engine failure on takeoff. Both crewmembers were relatively inexperienced in DC-93 flight operations. The FAA Principal Operations Inspector who was responsible for oversight of Midwest Express was inexperienced in FAR 121 turbojet air carrier operations. A "silent cockpit" philosophy was suggested by Midwest Express in response to certain emergency situations, although the concept was not approved by the FAA and was in conflict with approved emergency procedures.

[ANALYSIS] N1O0ME was certified, maintained, and equipped in accordance with applicable FAA regulations and approved procedures. The original airplane certification process had required demonstration of relevant handling qualities of the airplane, including conditions normally encountered in the event of sudden loss of thrust of either engine. The results of this investigation did not reveal any handling characteristics of the DC-9-14 which were inconsistent with the original standards for certification of the airplane. For example, the pilots who participated in the Safety Board's DC-9-14 flight demonstration deseribed the airplane's handling characteristics as docile, even after the sudden and complete loss of thrust from the right engine in a simulated takeoff/climb phase of flight. Consequently, the Safety Board concludes that the loss of control of the airplane was not directly attributable to the loss of thrust from the right engine. The analysis of this accident thus examined those factors which, in conjunction with, the failure of the airplane’s right engine, might have caused the pilots to lose control. Those factors included: ° The possibility that fragments of the right engine separated with sufficient energy and trajectory to cause critical damage to the airpiane’s flight control system; ~47T- ° The possibility of control system malfenction(s) which, in combination with a single or dual power loss, could have rendered the airplane uncontrollable; ° The possibility of a mechanical failure of the left engine, either related or unrelated to the failure of the right engine, which left the airpiane with insufficient thrust to maintain flight; and fe) The possibility of inappropriate flighterew response to the emergeney presented by the failure of the right engine. To resolve the factors which precipitated the loss of control, it was first necessary to examine the circumstances of the failure of the right engine. 2.2 ight ine Failure and Demage from Uneontained ine Parts The physical damage to the engine and the condition of the inlet fan biades and low pressure compressor blades indicated that the right engine had little or no rotation at impact with the ground. The sound spectrum examination of the CVRrecorded engine sounds also indicated that this engine lost rom very rapidiy after the engine failure. The hole in the high pressure compressor in the plane of rotation of the 9-10 Stage removable sleeve spacer and the damage to the compressor and spacer revealed conclusively that the spacer had ruptured in flight and that the spacer parts were not eontained by “he engine easing. The ejected spacer parts had ruptured the rear skirt intermediate .ase at the 11 to 1 o'clock position, leaving a 4-by 7-inch opening in the top of the case. The loss of the spacer and consequential Camage within the right engine caused a rapid deceleration and a complete loss of thrust from that engine.

[ANALYSIS] Witnesses were consistent in their deseriptions of the performance of the aircraft during the takeoff roll, rotation, liftoff, and initial climb. The observation by one DC-9 qualified pilot that the rotation to a takeoff attitude seemed abnormally abrupt was inconsistent with all other witness observations and was not corroborated by the FDR data. The aireraft performance cata, subsequent to the lftoff, was consistent with normal two-engine operation and disclosed evidence that all flight controls were functioning through the period of takeoff roll, rotation, and initial climb before the right engine failure. Thus, the Safety Board concludes that there were no operational irregularities of any consequence in those phases of the flight. Witnesses were consistent in reporting that their attention was attracted to the airplane Decause of one or more loud noises, described as “bangs,” and similar to "shotgun reports,” whieh occurred about the same time they saw flames and/or smoke from the right engine. The audible "bang," associated with an engine failure was confirmed by the CVR. Witnesses did not describe flame emitting from any part of the aircraft other than the right engine. Examination of the airplane confirmed that there was no in-flight fire other than thet contained within the right engine. The witnesses estimated that these events occurred about 300 feet a.g.l. it was determined that the right engine actually failed about 450 feet a.g.1. Sixteen witnesses reported that the aircraft seemed to decelerate following the right engine failure, consistent with FDR data. The deceleration of the airplane was eaused primarily by the loss of thrust from the right engine. The deceleration also was influenced by sideslip~induced drag and a reduction of left engine thrust. Reduced girspeed and increased G load made the airplane susceptible to accelerated stall. The presence of sideslip made the airplane susceptible to rolling motions. Correlation of the FDR, CVR, and air traffic control] data allowed the Safety Board to make several determinations regarding the flight. The airplane was climbing about 168 knots with a 50-foot per second rate of climb when the right engine suddenly failed. For 3 to 4 seconds, control was maintained with little change in heading, indicating that there was an initia: correct (left) rudder pedal application. Accelerometer data showed a reduction of normal G loads, indicating that the airplane's pitch attitude was lowered, apparently to reduce the rate of climb and to prevent a deterioration in speed. The left rudder pedal application and reduction of the sirplene's pitch attitude were consistent with the normal flight controi responses following loss of thrust from the right engine of a DC-3 airplane. About 4 seconds after the right engine failed, the airplane began to yaw rapidly to the right, as indicated on the FDR deta by the 20° ~-45- heading change from the third to the seventh second (after the right engine failure), while the radar data indieated that the airplane was continuing in a relatively straight track.

[ANALYSIS] About 4 seconds after the right engine failed, the airplane began to yaw rapidly to the right, as indicated on the FDR deta by the 20° ~-45- heading change from the third to the seventh second (after the right engine failure), while the radar data indieated that the airplane was continuing in a relatively straight track. The yaw rate was greater than that which would have occurred due tc a sudden toss of right engine thrust, or @ sudden release of the rudder pedal force used to compensate for the asymmetrical thrust. The Safety Board determined that the sideslip angle reached about 15° and that the total yaw reached 20° during this interval. The airplane heading deviated even ferther to the right and at a more rapid rate from the eighth to the ninth second, indicating that a large roil angle was developing. As the airplane started to descend, the normal acceleration forces increased rapidly. About 1 second later, 9.6 seconds after the right engine failure, the stali warning stickshaker act.vated when the normal acceleration indication was 1.5 G. Normal acceleration increased to about 1.8 G while the descending right turn continued, indicating that the airplane entered an accelerated stall at about 148 Knots. The airplane crashed about 5 seeonds later. (See figure 5.) The large heading changes which occurred later than the ninth second after the engine failed could not have occurred without the development of a large roll angle, in addition to right rudder deflection. Also, the ground track was not consistent with heading change due to roll angles, and the low normal accelerations (less than 1 G), which were recorded in this interval, wouid diminish the effects of roll angle on the heading change rate. Therefore, the Safety Board conchided that the sudden heading change which occurred before the eighth second after the right engine failure was caused by a yawing moment, rather than a rolling moment. The Safety Board believes that the configuration of flight 105 did not change after gear retraction. The flaps probably remained at 20° deployment until impact when they were driven farther downward to about 28° In other DC-9 aecidents, the Safety Board has found similar flap movement during the impact sequence. There was no evidence that the fiighterew initiated efforts to land the airplane. The investigation revealed <hat the flightcrew was medically and operationally qualified for the flight. They had received sufficient rest, and no evidence of adverse stress-related factors was found. Weather and air traffie control were not considered to be factors in the accident. N1O0ME was certified, maintained, and equipped in accordance with applicable FAA regulations and approved procedures. The original airplane certification process had required demonstration of relevant handling qualities of the airplane, including conditions normally encountered in the event of sudden loss of thrust of either engine. The results of this investigation did not reveal any handling characteristics of the DC-9-14 which were inconsistent with the original standards for certification of the airplane.

[ANALYSIS] The rapid deceleration would have resulted in a vestibular perception of downward pitching of the nose of the airplane. The Safety Board believes that more effective scanning of the flight and engine instruments by the pilots of fight 105 would have enabled them to maintain control of the airolane and to properly evaluate the powerplant anomalies. The failure of the first officer to respond to the captain's questions and the failure of the eaptain te maintain control of the airplane suggests that there was a breakdown in instrument sean dy Doth pilots in the critical seconds which followed the right engine failure. In view of the finding that the loss of control of the airplane probablv was eaused by the fiightcrew's impreper response ta the engine out emergency, the Safety Board examined several factors which could have contributed to the flig'terew's improper actions. 25.1 Fl ierew Training and DC-9 Qualification Airlines larger than Midwest Exoress, with many more vears of operating experience and larger pilot populations. upgrade pilots to captain Sased on demonstrated ability to accept the responsibilities of the position, sufficient seniority to suecessfully bid on the position, and completion of the required training. Midwest Express uses the same criteria; however, with a smaller pilot population, advancement to captain ean occur much sconer, as indicated by the advancement of the flight 105 pilots. Pilots at the more established airlines must have a great deal more senioritv and thus have more pilot experience in turbojet airplanes before captain upgrade because of the relatively slow growth of those airlines. Because of the DC-93s reletively small size in their fleets, it is typically the first turbojet airplane in which manv airline pilots upgrade to captain. Based on a sampling of recent upgrades to captain at two airlines, which the Safety Board oelieves are representative of carriers providing most of the scheduled passenger service in the United States, the Board determined that, Oy comparison, both pilots of flight 105 were relatively inexperienced in turbojet operations. For example, the experience level of recent DC-9 captain upgrades at the two airlines was: in excess of 10 vears’ seniority with the company, in excess of 10,009 hours total pilot experience including more than 7,500 turbojet hours as first officer, and generally served as a Slight engineer for more than 3 vears before upgrading to first officer. The Safety Board does not believe that much experience is essential for initial upgrade to captain of a DC-9; however, extra experience does provides a greater margin of safety to the traveling public. By contrast, the captain of fight 105 had been employed by Midwest Express for 12 months and had 606 hours of turbojet experience as a DC-9 first officer (no flizht engineer experience) ai the time of his captain upgrade. He had no turbojet or sweptwing airplane experience before being hired by Midwest Express. The first off icer of flight 105 had prcvious turbojet experience in the U.S.

[ANALYSIS] Any reduction in left engine thrust that occurred before stickshaker would have reduced the yawing of the airplane which occurred after the right engine failure. While possibly necessitating a foreed landing. a left engine power ioss should not have precipitated, or even contributed to, a loss of control of the airplane. However, the reduction in ieft engine power could have confused the crew. A detailed discussion of the Safety Board's analysis of the leit engine mechanical condition and operation is contained in appendix i. 25 Evaluation of Flightcrew Response The flight demonstration of a DC-9-14 airplane showed that with a sudden loss of right engine thrust at 170 knots, lateral and directional control could be maintained even if the cilot took no immediate action to deflect the rudder. Under thesc conditions, the airplane experienced about an 8° heading change and developed about 5° of sideslip within 4 seconds. About 30° of control wheel deflection, or 8° left deflection of the rudder, was required to maintain 4 wings-level attitude of the airplane. The flight demonstration was conducted with about 9,500 pounds of continuous thrust on the left engine (in-flight takeoff power). The sound spectrum examination disclosed that, on the accident airplane, the left engine thrust dropped from 19,750 pound (initially) to about 9,500 pounds after 2 to 3 seconds and to 5,500 pounds at the time of the loss of control. Because of the reduced asymmetric thrust, the yawing moment would have been reduced considerably on the accident airplane, similar to the demonstration airolene. Since there was no difficulty in compensating for the thrust asymmetry on the demonstration flight, the Safety Board concludes that the yawing moment should have been controllabie in the accident airplane. Since the airplane maintained its heading for the first 3 to 4 seconds after the rignt engine failed, it was concluced that the rudder was deflected properly to the left during that interval. However, based upon caleviations of the airplane's vawing response and resultant ground track for various rudder deflections and roll angles, the Safety Board Getermined that the large heading change and sideslip angle that developed after the first 4 seconds could not have been accomplished without a deflection of the rudder to the right, followed by a9 roll to the right 4 to 5 seconds later. Based upen the known performance of a DC~9-14, the closest duplication of the heading change which occurred on the accident flight (indicated by the FDR) would require the rudder to be deflected 6° to the left for about 3 seconds, followed by a rapid return of the rudder to neutral, then deflection of the rudder 12° to the riett about 3 seconds after the right engine failure. Returning the rudder to neutral and holding neutral rudder, after initially applying rudder to correct for differential engine thrust. wouid not have created the heading change rates which were indicated by the FDR data.

[ANALYSIS] S* PREDICTED NOMINAL GROUND IMPACT PC:NT “NN TAKEOFF FLIGHT P ESTIMATED POINT OF EJECTION ie) 250 FT 500 FT a a I LONGITUDINAL RUNWAY SCALE DISTANCE FROM END OF RUNWAY {{FT}} Figure 6.—-Caleulated nominal ground impact point and actual locations of engine debris adjacent to runway 19R. r Figure 7 shows that undefleeted parts are initialiy ejected in a direction approximated by the broomstick method; in this ease, 20° outbeard of vertical (away from the airplane fuselage) with high energy and speed. To have struck the Midwest Express fuselage, the parts would have had to have been deflected in excess of 65° and thus virtually all energy would have been lost. Study of the DC-9-14 control system revealed that all of its components which pass through the aft fuselage pass the engines below the cabin floor level. A were protected by multiple layers of aircraft structurc. Therefore, for ejected engine parts to nave struck and damaged any of these control system components, the ejected parts first would have to have been deflected more than 120° from their initial tangential ejection path and then would have to have penetrated and continued through engine cowling, engine pylon, fuselage, possibly fuselage supporting structure, the cabin floor, and possibly several intercostal flocr beams. Having reached the control system component(s), sufficient energy would have to have remained to disable the system components. The Safety Board believes that the possibility of ejected engine parts reaching internal control system components was extremely remote, if not impossible. The possibility of parts penetrating the fuselage at a point farther aft and damaging control components in the vertical fin would have been even more remote. Relatively low velocities would have been required for parts to have progressed in that Girection and to have struck the sirplane, while high energy would have been required to penetrcte the fuselage structure. Moreover, examination of the control system revealed redundancies which would have allowed the flightcrew to maintain full control of the airplane even if some control svstems had been disabled. Also, it was found that the rudder hydraulic actuator, which controls rudder movement by hydraulic pressure or by transferring control input to the aerodynamic tab, showed no evidence of preimpact damage. Additionally, the rudder power shutoff vaive was found with a bent control rod and discoloration, consistent with rudder hydraulic power on at ground impact.

[ANALYSIS] Also, it was found that the rudder hydraulic actuator, which controls rudder movement by hydraulic pressure or by transferring control input to the aerodynamic tab, showed no evidence of preimpact damage. Additionally, the rudder power shutoff vaive was found with a bent control rod and discoloration, consistent with rudder hydraulic power on at ground impact. The Safety Board also examined the possibility that the right engine cowl was biown open in flight or became distorted to such an extent that excessive drag was produced, affecting controllability of the aircraft. Aithough the right engine upper eowling was extensively damaged by impact forces, all four outboard iatches remained latched. There was no evidence to indicate that the right cowl had opened in flight. A recovered right cowl pieces which could be positively identified were found within the impact area. Although a small (2- by 2-inch) piece of metal whieh resembled cowl material was found near runway 19R, it was determined that each square foot of deformed cowling would produce drag equivalent to a reduction of engine thrust by 160 pounds--a minor factor. Based upon the small hole (4- by 77-inch) found in the right engine case, the absence of other case deformation (other than impact damages), and the characteristics of typical uncontained engine pieces ejected at high velocity, the Safety Board concludes that the cowling deformation probably was small and therefore caused very little additional drag following the right engine failure. 2.3 Flight Control System Failure or Malfunction The Safety Board considered the possibility of a flight control system failure or malfunction, unrelated to the right engine failure, that might have occurred simultaneously or nearly simultaneously with the right engine failure, and thet subsequently led to the loss of control. The Safety Board does not believe that such a failure or malfunction ceeurred for several reasons, ineluding those reasons cited previously regarding possible damage caused by the ri¢ht engine failure. In addition, an analysis of the contro! movements, which woukl have been required (commanded or etherwise) for the airplane to have maneuvered as indicated by the FDR, revealed that: 100% 80% DEBRIS 55% ENERGY (PERCENT) 0 ies ean en ee fi a I oococecocaccacdgeucesosazags ~*~ 06S, Figure 7.--Three-view drawing of a DC-3 series 10 airplane with various debris patterns and corresponding percentage of energy ioss. ~5i- (1) Rudder deflection to the left was required for the airplane to maintain heading for 4 seconds immediately after the right engine fejlure; (2) Rudder deflection to the right was required to cause the heading ehenge which occurred from the 4th to the 10th second after the right engine failure; (3) Elevator control was required to cause the pitcn-over and pull-up maneuvers which were documented by the FDR acceleration traces after the right engine failure; and (4) Aileron/spoiier deflection was required for the airplane to maintain the roll attitude in the presence of large sideslip angles which were docu mented.

[ANALYSIS] The Safety Board considered it plausible that the pilot's attention may have been directed to the movement of the left engine instruments after the right engine instruments became static. If he perceived that the left engine was the problem, he may have reacted to that perception by applying the rudder correction for a left engine failure rather than continuing with ‘he rudder deflection appropriate for a right engine failure. The attentiveness of the crew to flight instruments during the emergency anc a coordinated response to the indications were critical to maintaining control because a swept wing airplane, such as the DC-9, when in a sideslip, will tend to roll unless ecrrectiv2 action is taken. Either the sideslip must be reduced by appropriate rudder deflection or the lateral controls must be deflected to counter the rolling tendencies. The rolling tendency due to sideslip will increase as the lift on the wings increases. For a given G load (lift) and sideslip angle, a certain amount of lateral contro! deflection will be required to counter the roll. As the G load inereases, additional lateral control deflection will De required te counter the roll. in this case, elevatcr contre] input cxused the G load to increase from about 9.3 Gs to about 1.8. Gs from the 5th to the 11th second. In the presence of the sideciip angle and increasing acceleration load, the lateral control deflection would have tc be approximately doubled to maintain a constant dank angle. If the pilot established a lateral control deflection at the fourth to fifth second to compensate for the sideslip angle and then was not monitoring the roll attitude as the G load increased, the airplane would rofl further to the right. The data indicate that by the time the stickshaker came on, the airplane was in a significant roli attitude to the right and the positive G load was increasing. All of the above conditions should have deen evident to the flightcrew by reference to the flight instruments. The captain, had the individual ability and the responsibility to scan the instruments and to take corrective acticn. However, an appropriate coordinated flighterew response also would involve actions by the first officer to assist the captain in diagnosing and responding to the problem. The redundancy provided by the first officer is one of the basic tenets of cockpit resource management. 2.5.5 Cockpit Resource Management The investigation revealed that Midwest Express did not have a formal training rogram in cockpit resource management, which is also known as erew coordination. However, the Safety Board belNeves that with a low pilot to supervisor ratio. the airline could, and probably did, monitor closely the performance of its crewmembers, both as pilots and as individuals participating in the joint operation of flights.

[PROBABLE CAUSE] The evacuation was effective and completed within 1 1/2 minutes. PERERA ts Sosa ar SSE SN oe ee The emergency response to the accident was deficient in the limited number of ambulances dispatched to the site. Two passengers were reported to have been intoxicated at the time of the accident, and they could have adversely affected the evacuation. TREE RES, SET WS Sie tlc manle ee bron one 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of the accident was the captain's failure to stabilize the approach and his failure to discontinue the approach to a landing that was conducted at an excessive speed beyond the normal touchdown point on a wet runway. Contributing to the accident was the captain's failure tv optimally use the airplane decelerative devices. Also contributing to the accident was the lack of effective crew coordination during the approach. Contributing to the severity of the accident was the poor frictional quality of the last 1,500 feet of the runway and the obstruction presented by a concrete culvert located 318 feet beyond the departuro end of the runway. 4. RECOMMENDATIONS As a result of its investigation, the Safety Board made the following recommendations: —to the Federal Aviation Administration: Require airport manayters to repair areas and/or remove obstacles, such as concrete culverts, that are adjacent to airport operating areas. Such repairs should be performed at the earliest opportunity. (Class I], Priority Action) (A-87-107) issue an operations bulletin to principal operations inspectors of air carriers operating aircraft with flight attendants informing them of the need to cease providing alcohol to passengers who are in, or appear that they are about to be in, an intoxicated state. (Class Il, Priority Action) (A-87~-108) Issue an airworthiness directive for a one-time inspection of the seat pan roller assembly of the flight attendant seat, Trans Aero Industries, pert ne wes for evidence of fatigue cracks. (Class "I, Priority Action) A--87-109 During annual inspections of full certificate airports, emphasize the identification of deficient runway conditions and use approved frieticn-measuring devices to measure the dry runway coefficients of friction; encourage the airport operator to correct (or provide appropriate notice to users) runway conditions that do not meet the criteria recommended in Advisory Circular 150/5320-12A. (Class II, Priority Action) (A-87~110) : aving annual inspections of full certificate airports, verify that airport operations manuals address runway pavement inspection and maintenance criteria as recommended in Advisory Circular (AC) 150/5320-12A, and that airport operators are taking actions needed, including appropriate measurements of dry runway coefficients of friction with approved devices, to maintain runways tc the eriteria pecommended in AC 150/5320-12A, (Class Il, Priority Action) “87-111 ee aA ARES MOHELA Pa BORER SURE NE, SELINA TTA MPS RPS VANE WE AS ral eae Oa atta eNO RM Heath TOS ONE a Fe RS SoH ANA Salt teeny panes Goro saioed GA onsaeloet cate ~39~ —to the American Association of Airport Executives and the Airport Operators Council International, Ine.: Inform its membership of the elreumstances of the aircraft accident at Charlotte Douglas International Airport on October 25, 1986, and request its membership to repair areas and/or remove obstacles, such as concrete culverts, that are adjacent to airport operating areas.

[ANALYSIS] SETA HAN ERENT VINE SNA HARE ft Apter ert yee oN ec pre enema ayer noe aetna & departed the runway, as indicated by the reverted rubber marks found on the four main landing gear tires and the "steam clean" marks found on the departure end of the runway. Although runway frietlon was, according to FAA-recommended standards, not acceptable only near its departure end, the Safety Board concludes that the runway condition was not a primary cause of the accident because of the excessive speed of the airplane as it entered the last 1,500 feet of the suinway; but the poor friction did contribute to the severity of the accident. I Although the Safety Board concludes that the condition of runway 18L/36R did not contribute to the cause of the accident, the evidence indicates that the runway did not meet the maintenance standards recommended in FAA Advisory Cireular (AC) 150/5320-12A, dated July 11, 1986. The circular also indicates that the Charlotte Airport Authority did not comply with 14 CFR 139.83 regarding the prevention of ponding on runway pavement areas. The Safety Board belleves that as part of the FAA annual certification inspection of airports, such defects should be identified and corrected. - Currently, airports that are certificated under 14 CFR Part 139 are responsible for their own "self~inspection" program that, among other things, requires them to ensure that the airport pavement surface is adequately maintained. The Charlotte Airport Operations Manual (AOM) was examined subsequent io the accident.

[ANALYSIS] In light of the frictional deficiencies that were found on portions of runway 36R at Charlotte Airport, the Safety Board believes that the concepts at issue in Safety Recommendations A-82-153 and -154 still have considerable merit. However, ‘because the recent response indicates that FAA does not intend to take further action on these recommendations and because the Safety Board is issuing new safety recommendations concerning these issues, Safety Recommendations A-82-153 and -154 have been classified as "Closed~-Unacceptable/Superseded." Despite the FAA's position with regard to annual measurements of runway friction, the Safety Board also believes that the deteriorated condition of runway 36R at Charlotte Airport is indicative of failures on the part of the airport operator and the FAA inspectors to identify and correct other runway conditions that could adversely affect the safety of air carrier operations during inclement weather conditions. Further, the Safety Board believes that the recently revised AC 150/5320-12A should serve as a basis for an aggressive runway inspection and maintenance program. 2.6 Survival Aspects After it left the runway, the airplane struck and broke off the localizer antenna array from its frangible moorings. However, about 18 feet beyond the antenna was a concrete culvert which caused almost all the damage to the airplane and injuries to those who were injured. The Safety Board believes that the presence of the concrete culvert created a more destructive and severe accident than what it otherwise would have been without the culvert. : . . The Safety Board expressed {{ts concern about runway safety areas following a Texas International Airlines DC-9 accident at the Stapleton International Airport, Denver, Colorado on November 16, 1976. The airplane overran the runway during a rejected takeoff. Subsequent to the accident, the Safety Board recommended that the PAA: I A-77~-16 Amend 14 CFR 139.45 to require, after a reasonable date, that extended runway safety area criteria be applied retroactively to all certificated airports. At those airports which cannot meet the full criteria, the extended runway safety area should be as close to the full 1,000-foot length as possible. oo : : The FAA's initial response, dated July 11, 1977, stated that this recommendation would place an economic burden on airport operators. They did propose, however, an amendment to 14 CFR Part 139 that would require extended safety areas concurrently with construction of new airports, runways, and major runway extensions at existing airports. On October 23, 1985, the FAA published Notice of Proposed Rulemaking (NPRM) No. 85-22, "Revision of Airport Certification Rules," published at 50 FR 43094.

[ANALYSIS] The Safety Board belleves that as part of the FAA annual certification inspection of airports, such defects should be identified and corrected. - Currently, airports that are certificated under 14 CFR Part 139 are responsible for their own "self~inspection" program that, among other things, requires them to ensure that the airport pavement surface is adequately maintained. The Charlotte Airport Operations Manual (AOM) was examined subsequent io the accident. It stated that "the runways have been designed to provide 1 1/2 percent crown... all of the runways are grooved full length and width to facilitate runoff." Because of the deficiencies that were found in the condition of runway 36R (i.e, it did not have 1 1/2 percent crown in over half the length, the grooving was substantially collapsed in the last 1,500 feet, there were ruts (which were conducive to ponding) for almost the entire length, and the measured friction over the last 1,500 feet was substandard), the Safety Board believes that the airport operator failed to maintain the runway surface to standards specified in the AOM or to the criteria recommended in AC 150/5320~12A., Subsequent to the World Airways DC-10 overrun at Boston-Logan international Airport on January 23, 1982, 11/ the Safety Board recommended that the A: A-82~153 Use a mechanical friction measuring device to measure the dry runway coefficient of friction during annual certification inspections at full certificate airports and require that a Notice to Airmen (NOTAM) be issued when the coefficient of friction falls below the minimum value reflected in Advisory Circular 150/5320-12, Chapter 2. An82-154 Require that full certificate airports have a plan for periodic inspection of dry runway surface condition which includes friction measuring operations by airport personnel or by contracted services and which addresses the training and qualification of operators, calibration and maintenance of the equipment, and procedures for the use of the friction measuring equipment. 11/ Aircraft Accident Repurt--"World Airways, Ine., Flight 30H, MeDonnell Douglas DC- 10-30, Boston-Logan International Airport, Boston, Massachusetts, January 23, 1982" (NTSB- AAR- 82-15). “a aa PR GE SOE FRR ER RE NII rE RoR ROR Vaan Soom ON RAE peeene mean compe onan mee rs aos 7 Sern eee HO AREA EDEN BEN, RNASE MER: SEO ER AE MN PO I 1 -$4- On January 14, 1987, the FAA responded to these safety recommendations stating that "... the FAA does not believe that measuring dry runway coefficient of friction during certification inspections would be cost-effective nor would any ‘ ‘tnifleant safety improvement result" and indicated that no further action was contemplated. In light of the frictional deficiencies that were found on portions of runway 36R at Charlotte Airport, the Safety Board believes that the concepts at issue in Safety Recommendations A-82-153 and -154 still have considerable merit. --- Footnotes: [11/ Aircraft Accident Repurt--"World Airways, Ine., Flight 30H, MeDonnell Douglas DC- 10-30, Boston-Logan International Airport, Boston, Massachusetts, January 23, 1982" (NTSB- AAR- 82-15).]

[ANALYSIS] More important, perhaps, is the difficulty they may face when attempting to influence the pilot-in-command to reconsider and possibly alter a decision. Thus, it would have been very difficult, onee inside the final approach fix, for the first officer to suggest to the captain that the approach was not stabilized and, as a result, they should go around. Such a suggestion could, if presented inappropriately, distract the captain and could potentially endanger the safety of flight. As « result of its investigation of an airplane accident involving a Lockheed Electra L-188C in Reno, Nevada, on January 21, 1985, 10/ the Safety Board recommended that the FAA: A-86 =19 Provide to all operators, guidance on topics and training in cockpit _Tesource management so that operators can provide such training to their flighterew members, until such time as the FAA's formal study. of the topic {{s completed. 7 On December 19, 1986, the FAA informed the Safety Hoard that its study on cockpit resource management was expected to be completed in November 1987. As a result, the Safety Board has classified Safety Recommendation A-86-19 as "Open Acceptable Action" until it can review the results of the study. Until that time, the Safety Board reiterates Safety Recommendation A-86-19 and urges the FAA to provide guidance on cockpit resource management to all operators. It is hoped that operators will then implement such courses and provide training in the topic to all flighterew members. 2.5 ~ Runway Condition The Safety Board believes that (wo factors increased the severity of the accident: the lack of adequate runway friction in the final 1,800 feet of the runway and the location of the concrete culvert 18 feet beyond the localizer antenna array, which was itself, loca‘ed 300 feet beyond the departure end of the runway. The lack of acceptable friction in portions of the runway increased the severity of the accident because the airplane departed the runway at a higher speed than It probably would have had there been adequate grooving and drainage in the departure end of the runway. The evidence indicates that Pl 467 experienced hydroplaning before It 10/ Aircraft Accident Report--"Galaxy Airlines, Ine., Lockheed Electra L~188C, N5532, Reno, Nevada, January 21, 1985" (NTSB/AAR-86/01). I a Ar erent enen wenn estoy I en SAA RR AN RN PU NR tr OE OMEN Sy ONY Are! SETA HAN ERENT VINE SNA HARE ft Apter ert yee oN ec pre enema ayer noe aetna & departed the runway, as indicated by the reverted rubber marks found on the four main landing gear tires and the "steam clean" marks found on the departure end of the runway. --- Footnotes: [10/ Aircraft Accident Report--"Galaxy Airlines, Ine., Lockheed Electra L~188C, N5532, Reno, Nevada, January 21, 1985" (NTSB/AAR-86/01). | a]

[ANALYSIS] At the time of the accident, LaGuardia Airport was in instrument meteorological conditions due to an indefinite ceiling, 700 feet vertical visibility, and 3/4-mile prevailing visibility in light snow and fog. Although such conditions had been reported at LaGuardia since 2050, the Safety Board determined that the surface condition of runway 13/3! was acceptable for safe operations since the coefficient of friction and the depth of the wet snow were within acceptable operating limits. Plowing and sanding of the runways had been appropriately conducted and were continuing as needed. In addition, NOTAMs”? had been transmitted, or were currently being transmitted, that accurately described runway surface conditions at the time of the accident. The PIREP reporting a "nil" braking action on runway 4/22 resulted in the immediate and appropriate closure of that ninway. This resulted in increased delays and a longer holdover time for flight 405 after it had bern deiced at the gate. However, the Safety Board believes that the closure of runway 4/22 was an RY NY —_ \9Foushee, H.C., Lauber, J.K.. Baetae, M.M, and Acomb, D.S , 1986. Crew Faciors in Flight Operations HI: The Operational Significance of Exposure to Short-Haul Air Transpom Operalicas. NASA Technical Memorandum 88322, NASA-Ames Research Center, Moffett Field, Caiifonsia. oi'ce to Airmen. 47 operational necessity to ensure the safety of operations on that runway. This factor did contribute to the delays encountered by departing airplanes. The evidence gathered from the CVR and the FDR, as well as the statements of the first officer and passengers, revealed that after liftoff, the airplane could not transition to a positive climb angle. This situation indicated that the aerodynamic lifi-producing capability of the wings was degraded. There are numerous possible reasons for a loss of aerodynamic efficiency, such as an improper wine configuration, deployment of speedbrakes, and contamination or roughness of airfoil surfaces. There was no evidence that wing leading edge paint roughness or erosion/corrosion existed that could have degraded the airplane's performance. The fire patterns and damage to the speedbrakes showed that the speedbrakes were stowed before and during the accident sequence. The continuity of the airplane's flight control systems was examined and revealed no failure prior to impact. The six flap actuator jackscrews confinned that the flaps were set at 18 degrees, the proper configuration for takeoff from a contaminated runway. The wing and tail bleed air systems, including their seals, were intact, and the systems were found shut off. Therefore, the evidence indicates that there was no bleed air leakage that would have contributed to a loss of Sift during the takeoff attempt.

[ANALYSIS] Conversely, a greater rate of precipitation accumulation would have had the effect of reducing the holdover time of flight 405. The Safety Board believes that given the numerous variables and complexity of the problem, the specific amount of ice that accumulated on the aerodynamic surfaces of the airplane during the taxi phase is indeterminable. However, the Safety Board also believes that some contamination occurred in the 35 minutes following the second deicing and that this accumulation led to the control difficulty shortly after rotation. 2.9 Flightcrew Performance - Takeoff Procedure and Stall Recovery The Safety Board views the evidence as conclusive that the primary factor in this accident was the reduced performance of the wing due to ice contamination. Therefore, the Safety Board evaluated the extent to which the decisions of, and procedures used by, the flightcrew could have contributed to the accident. After arriving at the USAir gate following the landing at LaGuardia, both the captain and the first officer departed the airplane for short pertods, and both of them were aware that the weather conditions were conducive to the accumulation of frozen precipitation on the wings. Upon retuming to the airplane, neither of them performed a walkaround inspection or took any special actions to check the condition of the wing leading edge and upper surface. However, the airplane was subsequently deiced and the wing condition was purportedly checked by ground personnel which obviated the need for the crew to depart the airplane a second time for an extemal inspection. That the captain requested a second deicing after about a 20-minute delay indicated his concem about the continuing exposure to precipitation; the request was prudent and in accordance with USAir guidance. Following the second deicing, the flightcrew was most likely satisfied that the airplane was free of adhering contamination. The flightcrew was not aware of the exact delay that they would encounter before takeoff and their decision to leave the gate was reasonable. After taxiing, when it became evident that they would be delayed for a prolonged period, conversations between the crew showed that they were aware of and probably concemed about the risk of reaccumulating frozen contamination on the wing. Their awareness of this risk should have been heightened by the need to use the windshield wipers intermittently in combination with the freezing outside air temperature. When it became apparent that the delay would exceed 20 minutes, USAir guidance prescribes a careful examination of the airplane's surfaces. The first officer stated after the accident, and passengers confirmed, that he had tumed on the wing inspection light to view the wing on several occasions.

[ANALYSIS] The six flap actuator jackscrews confinned that the flaps were set at 18 degrees, the proper configuration for takeoff from a contaminated runway. The wing and tail bleed air systems, including their seals, were intact, and the systems were found shut off. Therefore, the evidence indicates that there was no bleed air leakage that would have contributed to a loss of Sift during the takeoff attempt. The evidence did not support improper wing configuration, airframe or system defects, or deployment of the speedbrakes as reasons for the loss of aerodynamic efficiency. Consequently, the analysis of this accident focused on the following: the weather affecting the flight; USAir's deicing procedures; industry airframe deicing practices; air traffic contro! aspects affecting the flight; USAir's takeoff and preflight procedures; and flightcrew qualifications and training. The dynamics of the airplane's impact with the ground, postaccident survivability, and crash/fire/rescue activities were also analyzed. 2.2 Prevailing Weather Conditions The Terminal Forecast for LaGuardia Airport, prepared by the National Weather Service (NWS), did not need to be updated at the time of the accident. The temperature recorded at the airport was below freezing, and wet snow was falling continuously for several hours prior to the accident. ‘Therefore, flight 405 was exposed to conditions that were conducive to airframe icing. 2.3 Flight Performance of USAir Flight 405 Aircraft headings and indicated air speeds obtained from the FDR were used to develop a time history of the airplane's ground track from the beginning of takeoff to the impact. Further, the acceleration during the takeoff, as derived from the air speed data, vas compared with the expected acceleration, as calculated by the manufacturer. The comparison of accelerations showed that the takeoff ground roll of flight 405 was normal. While ice contamination increases the drag produced by the wing, this effect is not signiftcant below the air speeds and high AOA associated with liftoff and initial climb. During flight 405's takeoff ground roll, wing AOA was near zero, and the air speed was relatively low. ‘The ground roll performance exhibited by the airplane was normal as would be expected with or without ice contamination on the wings. The Safety Board's evaluation of simulation data provided by Fokker for the conditions of the accident takeoff showed that the airplane without wing contamination would lift off about 2 seconds after the stait of rotation, assuming an average 3-degrees-per-second rotation rate. During the 2 seconds, the airplane would accelerate about 7 knots. Thus, with the start of the rotation at a pitch attitude of -! degree and a proper speed of 124 knots, the airplane would lift off as it reached 131 knots when the pitch attitude was about 5 degrees. The simulation data showed that the AOA would reach a peak of about 9 degrees as the airplane transitioned to the initial climb. With a stall AOA of 12 degrees in ground effect, the airplane, without wing contamination, would have at least a 3 degree-AOA stall margin during the transition to ciimb.

[FINDINGS] At the time of the accident, procedures for opening emergency door exits were inaccurately and incompletely displayed on USAir's F-28 passenger safety briefing cards, but they did not contribute to the fatalities in the accident. The locations of the dike, pump house, and ILS localizer ground plane antenna were within current FAA guidelines; however, the locations did not meet [CAO Annex 14 criteria. The overall emergency response was effective and contributed to the survivability of the airplane's occupants; however, the response by the emergency medical services personnel was inadequately coordinated, and the ambulance response times to the hospitals were excessive. The difficulties that the air traffic controller experienced with the emergency telephone system did not hinder or delay the ARFF response. 3.2 Probable Cause The National Transportation Safety Board detennines that the prebabie causes of this accident were the failure of the airline industry and the Federal Aviation Administration to provide flightcrews with procedurcs, requirements, and criteria compatible with departure delays in conditions conducive to airframe icing and the decision by the flightcrew to take off without positive assurance that the airplane's wings were free of ice accumulation after 35 minutes of exposure to precipitation following deicing. The ice contamination on the wings resulted in an aerodynamic stall and loss of control after liftoff. Contributing to the cause of the accident were the inappropriate procedures used by, and inadequate coordination between, the flightcrew that led to a takeoff rotation at a lower than prescribed air speed. 78 4. RECOMMENDATIONS As a result of this investigation, the National Transportation Safety Board makes the following recominendations: --to the Federal Aviation Administration: If gate holds are required to limit deicing fluid holdover time, encourage air traffic control (ATC) to initiate the gate holds as soon as a deicing operation begins rather than after delays have exceeded 15 minutes, as in the current air traffic control definition of gate hold. (Class Il, Prionty Action) (A-93-19) Where deicing operations are conducted away from the departure runway, report taxi delays in conditions conducive to airframe icing in increments that are less than {{5 minutes to provide more realistic and useful reports to dispatchers and flightcrews. (Class H, Priority Action) (A-93-20) Require that flight crewmembers and appropriate ground personnel responsible for the inspection of transport-category airplanes for wing contamination receive specific periodic training that will illustrate what contamination looks like and feels like on a wing and the amount of contamination that is detectable under different light conditions. (Class IT, Priority Action) (A-93-21) Study the effects on performance of swept-wing turbojet airplanes when specific amounts of air speed are added to the computed roiation speed (delayed rotation) during takeoffs when wing contamination is possible. (Class HN, Priority Action) (A-93-22) Require Fokker to detemnine how takeoff performance and stall margin would be affected by using a lower initial target pitch attitude on F-28 and F-100 airplanes in the event that undetected upper wing ice contamination is present, and change the nommnal!

[ANALYSIS] The Safety Board urges the Port Authority to continue this initiative and replace the two other pump houses, which are adjacent to runway 13/31, with buried installations. Replacement of the FAA ILS localizer ground plane antenna has already been accomplished; however, the Safety Board found that the antenna is of a simit: nonfrangible design as tlic ofigitial. The FAA Generai Enginecr, Office of Airpon Safety and Standards, testified that because of the unique location and design of the antenna, it was not technically feasible to make it frangible. ‘The Safety Board urges the FAA to conduct research on the frangibility of the antenna and to replace the current ILS localizer ground plane antenna with one that can function properly and is a less hazardous obstruction. 34F 50m 3.3.3. "A strip including a precision approach runway shall, wherever practicable. extend laterally to a distance of at feast: - £50 m (approximately 411 feet) where the code number is 3.07 4." 72 2.19 Airport Rescue and Fire Fighting Effectiveness 2.10.1 Communications The Safety Board concludes that the difficulties the ATC controller experienced with the emergency conference line did not delay cr hinder the emergency response because ARFF personnel heard the controller's first transmission. However, the Safety Board believes that a potential for a breakdown in communications exists until the deficiencies in the system are corrected. The Port Authority should expedite the replacement of the emergency telephone system. 2.10.2 Medical Response The Safety Board believes that factors contributing to the delay in transporting the eight passengers and one cabin crewmember who sustained serious injuries included the following: poor weather/road conditions; confusion in locating and treating a number of victims who had been transported by airline personnel to various locations around the airport; and the EMS failure to maintain continuous and close communication with the Incident Commander at the command post during triage operations. The Safety Board understands that during mass casualty incidents, the on-site treatment of victims by EMS personnel places first priority on medically stabilizing the injured prior to transporting them. However, seriously injured passengers were still arriving at area hospitals at 0015. Following the accident, sufficient resources were available to have stabilized and transported the injured more expeditiously. The Safety Board encourages the New York City Health and Hospitals Corporation and the Emergency Medical Service to review, in depth, and in concert with other New York City emergency response agencies, their response to the crash of USAir flight 405. The Safety Board believes that these services should continue to seek ways to improve coordination and to reduce the time required to transport injured persons to hospitals from LaGuardia Airport. The Safety Board also noted that victims who were removed from the water during the initial stages of the emergency response, and who lacked visible vital signs, such as pulse, and respiration, were categorized as deceased and that no attempts were made to resuscitate them.

[ANALYSIS] The_airplane's exit markings were in accordance with 14 CFR 25.811. Although none of the exits were used in this accident because survivors evacuated through openings in the fuselage, passengers must be provided adequate information so that they can open emergency exits. The Safety Board is concemed that FAA surveillance did not identify the inaccurate and incomplete information shown on the passenger safety briefing cards. 2.9.2 Runway Area Obstructions The Safety Board is concemed about the location of nonfrangible obstnictions in the vicinity of runway 13/31 that significantly contributed to the severily of damage. The locations of the dike, the ILS localizer ground plane antenna, and the pump house met the current FAA criteria for frangibility since both 71 structures and the dike were just outside the 500-foot runway sa:ety area. However, AC 150/5300-13, Airport Design, Appendix 8, par. 4 states: “The ROFA (Runway Object Free Atea) is a result of an agreement that a minimum 400-foot (120 m) separation from runway centerline is required for equipment shelters, other than localizer equipment shelters. Also, ICAO Annex 14, AERODROMES, Volume I Aerodrome Design and Operations, 8.6.1. states: “Unless its function requires it to be there for air navigation purposes, no equipment or installation shall be: a) ona runway strip,-4 a runway end safety area, a taxiway strip or within the distances specified in Table 3-1, column 11, if it would endanger an aircraft...” Although the localizer ground plane antenna, pump house, and dike did not meet the criteria of AC 150/5300-13, Appendix 8 or the ICAO 8.6.1., the Safety Board understands the difficulties that LaGuardia faces in that regard, since the airport is physically restrained by size, location, and water boundaries. The Port Authority Assistant Director of Aviation testified that the pump house, which was destroyed in the accident, was to be replaced by a newer underground pump house, which was not technically feasible at the time of the construction of the original pump house(s). The Salety Board is pleased that the Port Authority took this initiative to further improve the safety of the environment around runway 13/31. The Safety Board urges the Port Authority to continue this initiative and replace the two other pump houses, which are adjacent to runway 13/31, with buried installations. Replacement of the FAA ILS localizer ground plane antenna has already been accomplished; however, the Safety Board found that the antenna is of a simit: nonfrangible design as tlic ofigitial.

[FINDINGS] There was no taxiway guidance sign at the intersection of taxiway W-1 and the east-west taxiway. Operators of airports certificated under 14 CFR Part 139 are not required to place standardized signs at each taxiway/runway and taxiway intersection. Runway signs should be sufficiently different in design from taxiway signs so thet they alert the operators of ail surface vehicles and airplanes of the nature of the intersection. 17. Lighted runway/taxiway signs should be inspected daily to ensure their operability and maintained as required. 3.2 Probable Cause The National Transportation Safety Board determines that the probable causes of ths accident were the failure of the pilot of Korean Air Lines Flight 084 to follow accepted procedures during taxi, which caused him to become disoriented while selacting the runway; the failure of the pilot to use the compass to confirin his position; and the aceision of the pilot to take off when he was unsure that the aireraft was positioned on the correct runway. Contributing to the accident was the fog, which reduced visibility to a point that the pilot could not ascertain his position visually and the control tower personnel could not assist the pilet. Also contributing to the accident was a lack of gible taxiway and runway signs at several intersections passed by Flight 084 while it was a oe SNH A ROPE 9 “ SORRY BENE LE AER se SEES gy I ERRATA ET Rp cor Fete ae: eee seer fe vad . aed ‘ f > 4. RECOMMENDATIONS I As 4 result of this accident investigation, the National Transportation Safety Board recommended that the Federal Aviation Administration: Require that airports certificated for air carrier operations install signs at all runway and taxiway entrances, exits, and intersections that indicate the identity of the runway or taxiway. (Class ii, Priority Action) (A-84-98) Require that the graphics on taxi:vay/runway identification signa be standardized and of sufficient size to enable them to be legible to aircraft crewmembers in all meteorological conditions in which air carrier operations are authorized. (Class i, Priority Action) (A~-84--99) Require that airport operators inspect and tmatntain the lights illuminating airport taxiway/runway identification sigs as part of the cay rita inupection requizements. (Cless I, Friority Action) A-84-100 Require at all airports certificated for air carrier operations that uniform signs be installed which are classified by function (e.g., runway entrance, runway exit, taxiway intersection) with each function having a unique shape, color, end/or size so that runway entrance signs are distinguishaole from all other advisory siens on airport property. (Class Ii, Priority Action) (A~84-101) Require that air carriers incorporate in training of their crewmembers procecures and responsibilities during ground operations in restricted visibility conditions, to enable them to operate sefely in such conditions. (Class I, Priority Action) (A-84-102) - BY THE NATIONAL TRANSPORTATION SAFETY BOARD — /s/ JIM BURNETT Chairman /s/ PATRICIA A, GOLDMAN Vice Chairman /s/ GH, PATRICK BURSLEY Member VERNON L, GROSE Member August 9, 1984 oh emedeal Fey OSU Se ne ara ed ie sere Preceding page blank = -27- APPENDIXES APPENDIX A INVESTIGATION AND HEARING

[ANALYSIS] The FAA should require under 14 CFR Part 139 that airport operators place appropriate runway or taxiway signs at cach intersection along alrport taxiways to designate either the intersecting taxiway or runway. The crew of KAL 084 did not Indicste in their statements that they saw the fully illuminated sign designating runway 6L/24R. Several factors may have contributed to the failure of the crew of KAL 084 to notice this sign, even though it was fully {{iluminated. The sign was dirty, which reduced the contrast between its background and lettering. Since the airport surfaces were obscured partially by snow, frost, and ice, the crew was looking intently for ground markings. Moreover, the Visibility was restricted, which further Hmited the crew's ability to see the sign, particularly since the location of the DC-10 cockpit is about 30 feat above the ground increases the slant range from cocknit to guidance signs placed aside taxiways and runways. ~-23- Contributing to the crew's failure to notice the runway sign was that, despite the different purposes that the runway and taxiway signs serve, the signs had common shape, color, and dimensional characteristics, The runway and taxiway signs had identical amber backgrounds with black lettering. The characters on the signs were identically sized. The signs, which were the same height, differeg only in their width according to the number of characters on the sign. The Safety Board is concerned that in similar situations other flighterews or vehicle operators could inadvertently enter an active runway. Runway and taxiway intersection signs should reflect, in their sizes, shapes, colors, and dimensions, the particular route they mari; a sign identifying a taxiway intersection should have a different appearance from a sign identifying a runway, and these signs should then de installed at airports certificated under 14 CFR Part 139. 2.7 Runway incursions The October 1978 ASRS article concerning human factors associated with runway incursions, as well as the three subsequent acciderits described earlier, substantiates problems and causal elements similar to those in this accident. While the December 19, 1983, accident at Anchorare and the collision at Sioux. Falls involved air traffic contro}}, the accident at: Madrid was similar to this accident. The Aviaco Airlines DC-8 pilot did not taxi as instructed at Barajas Airport during restricted visibility conditions. While the KAL 084 crewmembers did not ignore tower instructions, the factors of crewmember disorientation, cockpit coordination, and pilot technique cited in the ASRS article were evident in this accident. Flighterews must be especially vigilant during taxi, hold, and takeoff operations and must make extraordinary efforts if needed to stay aware of their position on the airport at all times. Crew coordination procedures should be enhanced and particular alertness should be practiced when visibility is reduced by inclement weather. I }} 3. CONCLUSIONS

[ANALYSIS] I The primary sources of information that are ordinarily available to crewmembers for guicance on airport surfaces were either partially or completely unavailable to the crew of KAL 084. At nighttime or under limited visibility conditions, crewmembers rely on runway surface markings such as taxiway lines and runway numbers, taxiway and runway lights, and runway and taxiway signs to provide them with information concerning their location on the airport. If the visibility is adequate, or If the airport is equipped with ASDE, ground controllers can assist the aircraft crewmembers by providing information on their location. The flighterew of KAL 084 operated essentially without external information to assist them while taxiing since the visibility was restricted and the airport did not have ASDE. a 2.5 _ KAL 084 Fightcrew Medical and Behavicral Factors The meiical examination of the KAI, 084 crewmembers immediately after the accident and the toxicological testing of bicod samples did not reveal any physiological condition which might have affected their performance. Each crewmember was well rested before the flight, having been «ff ducy for over 26 hours prior to the scheduled departure time. The crewmembers were housed in facilities operated by Korean Alr Lines for employees laying over {{n Anchorage to insure that crewmembers rest in an undisturbed environment with: Korean food and a familiar atmosphere. The performance of the crewmembers cannot be attributed to fatigue resulting from excessive duty time or te Stress created by unfarmilin’ surroundings. Similarly, interviews with the crew and their statements did not revwal any significant event in their lives thet may have caused them stress or tension or affatted their decisionmaking abilities. The flight wes not significantly delayed, nor was the crew facing an imminent deadline for completing the — flight, such ae deteriorating weather at destination, curfews, or excessive duty time. — From the response of the captain of KAI. 084 to questioning, the Safety Board could not determine why a: axperienced crew, such as this crew, did not werify whether they were on the correct runway by checking their heading instruments. The Safety Board could not find any factcr which may have adversely affected the crew's vision, coordination, or decisionmaling capabilities to determine that their heading was 80° from the correst runway bearing. ‘The failure of the crew to verity the runway heading may — indicate that the initial or rocurrant training the crew received or the op?rating procedures established for KAI, crewmembers are deficient. It may be chat verification of runway heading is such a rudimentary procedure that the air ecezrier believed that specialized training was not necessary.

[ANALYSIS] The Safety Board examined several of the runway and taxiway signs at the airport to determine if all of the avaiiabie sources of ground location information external to the airplane were adequately presented to the KAL 084 crew. The KAL airplane passed four signs identifying “unways and taxiways along the route that the Board believes it took while taxiing. One of the four signs, the sign designating taxiway N-1, was not equipped for electrical illumination. At night In restricted visibility conditions when additional guidance is most needed, such as existed at the time of this crash, this sign would provide no information or guidance to flighterews, Another of the four signs was only partially illuminated, because only three of its seven lights were operating at the time of the accident. The other two signs, which identified runway 14 and runway OL/24R, were iiluminated. Airports certificated under 14 CFR Part 139 are not required to have taxiway/runway guidance signs installed. However, if the signs are instalied, 14 CFR 139.47(b) requires that. the operator "must show that any guidance signs installed at the airport are in operable condition." Por each airport certificated under 14 CFR Part 139, the FAA approves an Airport Operations Manua! (AOM), which, in part, lists key elements of the airport, such as runway lights, that are required to be inspected daily to ensure that they are in operable condition. For many airports, including Anchorage International, the approved AOM does not. include guidance signs in the list of key elements. Therefore, although 14 CFR 139.47(o) requires that the signs be in operable condition, the FAA has not euppiied guidance to the airport operators on how or when this requirement will be met. The Safety Board believes that as KAL 084 taxied along texiway W-1, the crew thought that they were on the east-west taxiway, and that when they crossed the east-west taxiway, they thought it was the north-south taxiway and continue~ to what they believed was runway 32 but was instead runway 24R. There were no Sig..3 along this ground path to indieate, first, that the taxiway they had entered was W~1 and, second, that the first intersection they then crossed was the east-west taxiway. The crew of KAI, ignate either the taxiway they were on or axied to the intersection, of taxiway W~-1 signs have been installed at both intersections to te the intersecting taxiways. The FAA should require under 14 CFR Part 139 that airport operators place appropriate runway or taxiway signs at cach intersection along alrport taxiways to designate either the intersecting taxiway or runway. The crew of KAL 084 did not Indicste in their statements that they saw the fully illuminated sign designating runway 6L/24R.

[ANALYSIS] All of the involved flighterew members held current medical certificates. 2.2 Weather (art ee mate fergie ede The surface visibility at Ancharage International Airport was restricted, as evidenced by the 1350 surface observation which reported 1/8 mile visibility and the 1415 cbservation which reported 1/16 mile visibility. The local controller advised SCA 59 that the RVR was 1,000 feet at 1344:18, and the RVR did not improve to 1,800 feet until 1405:28, at which time SCA 59 was cleared to taxi into position and to hold on runway 6L. An RVR of 1,800 feet was the minimum takeoff visibility for the pilot of SCA 59. The captain of KAL 084 stated that, after he began taxiing from the parking ramp, he could see the yellow taxi lines "very dimly through the heavy ice fog." He devecribed the visibility as "so poor thay it was difficult to see the taxiway markings." After the accident, the first officer of KAL 084 concluded a written statement as fellows! "It seems that I lost my sense of direction due to the heavy ice fog, and I confuse’ the east-west taxiway with the north-south taxiway.” The testricted visibility caused the flighterew of KAL 084 to experience difficulties while operating on the taxiways and runways at Anchorage international Airport and adversely affected their operational performance. 2.3 Collision Analyals According to applicable performance charts, based on the estimated TOGW of $02,790 pounds, the temperature of 15 degrees F., and the field elevation of 144 feet, the departure runway length required for KAL 084 was 8,150 feet. The distance from the intersection of runway 6L/24R and texiway W-1, where KAL 034 began its takeoff roll, 10 the departure end of runway 248 is 2,400 feet. Based on these data, it can be concluded that the atiompted takeoff by the KAL 084 flightcrew would not have been successtul even i! their takeoff run had not been interrupted by the collision with SCA 59. _ -KAL 084 was equipped with three main gea.s, one being a centered body gear, — Given the dimensions of both airplanes, and the impact marks on SCA 59, it appears that the nore gear of KAL 084 struck SCA 59 on the right windscreen at the top and grazed the skin of the right fuselage over the cockpit, missed the remainder of the fuselage, and struck the vertical stabilizer. As the captain of KAL 064 turned left to miss SCA 59, the main body gear swung to the right and struck the left wing of SCA 59, knocking the wing off outboard of the engine nacelle.

[ANALYSIS] For example, in January 2014, a Boeing 737 landed at the wrong airport in Branson, Missouri, in night VMC. The flight crew expected that the visually identified airport and runway were the intended destination and did not reference cockpit displays to verify the airport and runway. As a result, the airplane landed on runway 12 at M. Graham Clark Downtown Airport instead of runway 14 at Branson Airport. Also, in November 2013, a Boeing 747 landed at the wrong airport in Wichita, Kansas, in night VMC due to the flight crew’s expectation that the observed runway lights were from the intended landing runway at McConnell Air Force Base. Instead, the airplane landed at Colonel James Jabara Airport on a runway that was one-half the length of the intended landing runway. For both of these cases, cues that indicated the flight crew’s mistaken perception were available; however, those cues were not effectively used because the crewmembers’ expectation bias outweighed the available conflicting cues.101 For this incident, lighting aids generally associated with runways were not present on taxiway C. Specifically, although the flight crew perceived the taxiway to be the intended runway, the taxiway did not have a precision approach path indicator, touchdown zone lights, full-length edge lights, and approach lights.102 However, the absence of these normally conspicuous features of a runway would have been difficult for the flight crewmembers to recognize because of their expectation bias and the inherent difficulty detecting omissions in the environment (the latter of which could have been mitigated if the flight crew had briefed the runway 28L closure). In addition, features present along taxiway C were inconsistent with it being a runway. For example, although the presence of centerline lights along the full surface length was a cue that was consistent 101 For more information, see NTSB incident numbers DCA14IA037 and DCA14IA016, respectively. 102 Air Canada’s stabilized approach criteria for a visual approach included vertical tracking on approximately a 3° glidepath and using a visual approach slope indicator. Postincident interviews and airplane track data suggested that the captain used the precision approach path indicator located to the left of runway 28R for glidepath information, which was intended for airplanes approaching that runway. The availability of this glidepath information while the airplane was aligned with taxiway C would have supported the flight crew’s expectation that the airplane was aligned with runway 28R. NTSB Aircraft Incident Report 51 with a runway, the taxiway centerline lights were green, as shown in figure 4.103 (Runway centerline lights are white.) Also, flashing yellow in-pavement guard lights were present on taxiway C (also shown in figure 4), which would not have been present on a runway surface because the guard lights were designed to prevent a taxiing airplane from crossing onto a runway. During postincident interviews, the flight crewmembers recalled seeing specific color cues, including the green taxiway centerline lights.104 However, the flight crew continued the approach despite this conflicting cue.

[CONCLUSIONS > FINDINGS] NTSB Aircraft Incident Report 67 9. Although the notice to airmen about the runway 28L closure appeared in the flight release and the aircraft communication addressing and reporting system message that were provided to the flight crew, the presentation of the information did not effectively convey the importance of the runway closure information and promote flight crew review and retention. 10. The cues available to the flight crewmembers to indicate that the airplane was aligned with a taxiway were not sufficient to overcome their belief, as a result of expectation bias, that the taxiway was the intended landing runway. 11. Multiple salient cues of the surface misalignment were present as the airplane approached the airport seawall, and one or more of these cues likely triggered the captain’s initiation of a go-around, which reportedly occurred simultaneously with the first officer’s call for a go-around. 12. The captain and the first officer were fatigued during the incident flight due to the number of hours that they had been continuously awake and circadian disruption, which likely contributed to the crewmembers’ misidentification of the intended landing surface, their ongoing expectation bias, and their delayed decision to go around. 13. Current Canadian regulations do not, in some circumstances, allow for sufficient rest for reserve pilots, which can result in these pilots flying in a fatigued state during their window of circadian low. 14. Flight safety would be enhanced if airplanes landing at primary airports within Class B and Class C airspace were equipped with a cockpit system that provided flight crews with positional awareness information that is independent of, and dissimilar from, the current instrument landing system backup capability for navigating to a runway. 15. Although the investigation into this incident identified significant safety issues, cockpit voice recorder information, had it been available, could have provided direct evidence regarding the flight crew’s approach preparation, cockpit coordination, perception of the airport environment, and decision-making. 16. Once the flight crewmembers perceived lights on the runway, they decided to contact the controller to ask about the lights; however, their query was delayed because of congestion on the tower frequency, which reduced the time available for the crewmembers to reconcile their confusion about the lights with the controller’s confirmation that the runway was clear. 17. Although the use of line up and wait (LUAW) procedures during single-person air traffic control operations was not a factor in this incident, the tower controllers should have delayed consolidating local and non-local control positions until LUAW procedures were no longer needed. NTSB Aircraft Incident Report 68 18.

[ANALYSIS] The flight crewmembers’ training records indicated no issues with identifying airport surfaces, flying stabilized approaches, and flying visual approaches. The incident occurred in night VMC, and no evidence indicated any obstructions or glare in the cockpit that would have affected the flight crew’s view outside of the cockpit windows. However, the flight crewmembers were unable to identify the runway 28R surface (despite the presence of approach and runway lighting) and instead aligned the airplane with parallel taxiway C. Also, neither crewmember recognized that the airplane was not aligned with the intended landing runway until the airplane was over the airport surface, at which time the flight crew initiated a low-altitude go-around. Sections 2.3.1 through 2.3.3 discuss reasons for the flight crew’s alignment error and the factors that led to the eventual recognition of this error, and section 2.3.4 discusses the mitigation of such errors. 2.3.1 Flight Crew Awareness of Runway Closure The flight crew had opportunities before the approach to learn about the runway 28L closure. The first opportunity occurred before the flight when the crewmembers received the flight release. Both crewmembers stated that they reviewed NOTAMs in the flight release. However, the first officer stated that he could not recall reviewing the specific NOTAM that addressed the runway 28L closure. Also, even though the captain stated that he saw the runway closure information, his actions in misaligning the airplane demonstrated that he did not recall that information when it was needed, and he thought that runway 28R was runway 28L. The second opportunity occurred in flight during the crewmembers’ preparations for the approach to runway 28R. Both crewmembers recalled reviewing ATIS information Quebec, which they received via an ACARS message in the cockpit, but neither crewmember recalled reviewing the specific NOTAM in the ATIS information that described the runway 28L closure. Because the flight crewmembers either did not review or could not recall the information about the runway 28L closure, they expected to see two parallel runways while on approach to SFO and further expected that they would need to fly the approach to the right-side surface.95 The flight crew’s recent experience flying into SFO would have reinforced these expectations. For example, when the first officer flew into SFO 2 nights before the incident, the airplane used for that flight landed on runway 28R at 2305, which was about 18 minutes before runway 28L was closed. Also, the captain stated that he had never seen runway 28L “dark” and that none of his previous landings at SFO occurred when a runway was closed. A runway closure marker with a flashing white “X” was placed at the threshold of runway 28L to indicate the runway closure.

[ANALYSIS] The flight crew was properly certificated and qualified in accordance with Canadian regulations, Air Canada requirements, and 14 CFR Part 129. Flight crew medical conditions. The flight crew held valid and current Canadian medical certificates. The captain and the first officer reported no medications, medical conditions, or sleep disorders during their required medical examinations and during postincident interviews. Airplane mechanical conditions. The airplane was properly certificated, equipped, and maintained in accordance with Canadian regulations and 14 CFR Part 129. No evidence indicated any structural, engine, or system failures. Airport lighting. Runway 28R (the intended landing runway for the flight) and parallel taxiway C at SFO met the requirements for lighting in 14 CFR Part 139. The runway 28R approach lights were on as the incident airplane approached SFO. Title 14 CFR 139.311 required taxiways to have one of four types of lighting systems, one of which was centerline lights, which were present and illuminated on taxiway C. The runway and taxiway centerline lights were set at step 1 (out of 5), which was appropriate for the weather conditions on the night of the incident. NTSB Aircraft Incident Report 40 Thus, the NTSB concludes that none of the following were factors in this incident: (1) flight crew qualifications, which were in accordance with Canadian and US regulations; (2) flight crew medical conditions; (3) airplane mechanical conditions; and (4) airport lighting, which met US regulations. 2.2 Incident Sequence 2.2.1 Notification of Runway 28L Status SFO runway 28L was scheduled to close at 2300 on the night of the incident due to construction, and a NOTAM in the flight release for ACA759 provided information about the closure. The captain stated that he and the first officer discussed the runway 28L closure while at the departure gate but that they did not place much emphasis on this information because the captain thought that the flight would land before the runway would be closed.84 (The NTSB notes that the flight was originally scheduled to land 3 minutes after the runway 28L closure.) However, the airplane pushed back from the gate 30 minutes late due to the delayed arrival at YYZ of the airplane to be used for the flight and took off about 49 minutes later than the scheduled takeoff time. Even though the flight crewmembers knew, during their preflight preparations, about ACA759’s delayed departure, no evidence indicated that the crewmembers reconsidered the importance of the NOTAM information at that time or as the airplane approached SFO.

[CONCLUSIONS > FINDINGS] Although the use of line up and wait (LUAW) procedures during single-person air traffic control operations was not a factor in this incident, the tower controllers should have delayed consolidating local and non-local control positions until LUAW procedures were no longer needed. NTSB Aircraft Incident Report 68 18. If an airplane were to align with a taxiway, an automated airport surface detection equipment alert would assist controllers in identifying and preventing a potential taxiway landing as well as a potential collision with aircraft, vehicles, or objects that are positioned along taxiways. 19. Increased conspicuity of runway closure markers, especially those used in parallel runway configurations, could help prevent runway misidentification by flight crews while on approach to an airport. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this incident was the flight crew’s misidentification of taxiway C as the intended landing runway, which resulted from the crewmembers’ lack of awareness of the parallel runway closure due to their ineffective review of notice to airmen (NOTAM) information before the flight and during the approach briefing. Contributing to the incident were (1) the flight crew’s failure to tune the instrument landing system frequency for backup lateral guidance, expectation bias, fatigue due to circadian disruption and length of continued wakefulness, and breakdowns in crew resource management and (2) Air Canada’s ineffective presentation of approach procedure and NOTAM information. NTSB Aircraft Incident Report 69

[ANALYSIS] The runway was dark because its lights were out of service, and the airplane landed on the taxiway that was parallel to the runway despite the illumination of blue taxiway edge lights, which the flight crew perceived as dim white runway edge lights. For more information, see NTSB incident number DCA18IA081. 104 Taxiway lighting at YYZ, where the flight crew was based, was similar to the lighting on SFO taxiway C; both airports had green taxiway centerline lights and blue taxiway edge lights. 105 The airplanes’ wingtip navigation lights were red (left wing) and green (right wing), and the runway edge lighting was white. 106 The DAL521 captain also stated that, because the lights from the airplanes located on taxiway C were in a straight line, the airplane lights could have been perceived as centerline lights, which could confuse a flight crew. NTSB Aircraft Incident Report 52 immediately recognize that the surface ahead was not the intended landing runway. When the first officer contacted the controller, the airplane was about 4,000 ft (0.66 nm) from the airport seawall. As ACA759 continued to approach the airport seawall, the flight crew of the second airplane on taxiway C, PAL115, saw that ACA759 was lined up with taxiway C, and the PAL115 crew turned on that airplane’s landing lights to alert the ACA759 crew. Video information showed that PAL115’s landing lights illuminated the surface in front of that airplane as well as the tail and side of the first airplane on taxiway C, UAL1. About this time, the UAL1 captain made the first of two transmissions, on the tower frequency, about the position of an airplane above the airport surface; these transmissions were also available cues for the incident flight crew to recognize ACA759’s misalignment.107 In addition, about this time, ACA759 descended to an altitude at which its landing lights would have illuminated the environment below. The appearance of airplanes on the surface (especially given that the controller had just advised the flight crew that runway 28R was clear) and the lack of runway markings on the surface should have been additional cues indicating that the airplane was not aligned with runway 28R. During postincident interviews, the captain and the first officer were unable to identify a specific triggering factor in the environment that led to the decision to initiate and call for, respectively, a go-around. However, all of the cues mentioned above occurred within 6 seconds before the initiation of the go-around (2355:59 to 2356:05). That period is consistent with the time for pilots to recognize a cue, make a decision, and execute an action.

[ANALYSIS] Given the current specifications for the runway closure marker, these flight crews likely did not see the “X” because their airplanes were not aligned with runway 28L. This situation highlights the need for a surface-based system with conspicuous and unambiguous visual cues that clearly indicate when a runway is closed, even if an approaching airplane is not aligned with the closed runway. Such a system would provide redundancy in case a flight crew does not review or retain runway closure information presented in NOTAMs or ATIS broadcasts/messages. Such a system would also be especially critical in situations involving the closure of a runway at an airport with parallel landing and taxiway surfaces (such as SFO). As previously discussed, cockpit mitigations (such as the use of navigational aids to back up a visual approach to the intended landing runway, as used by the DAL521 flight crew, and a system to provide additional positional awareness information) and ATC systems to alert controllers about a potential runway misalignment or taxiway alignment can provide multiple layers of protection to prevent a flight crew surface alignment error. In addition, when closed runways are marked conspicuously, the view outside the cockpit (during VMC) of an airport environment can assure flight crews that an airplane is correctly aligned with a parallel landing runway. If the incident flight crewmembers had observed the runway 28L closure marker early in the approach, their mistaken perception of the airport environment and alignment with taxiway C might not have occurred. Further, if the runway 28L closure marker had captured the incident flight crew’s attention later in the approach, that information might have been sufficient for the crewmembers to detect their mistaken perception and respond to the situation before the airplane reached the seawall. The NTSB is not aware of any operational or human factors research to improve the conspicuity of L-893 runway closure markers during the 31 years since the FAA issued a technical report about its research on runway closure markers (other than subsequent research on the use of light-emitting diode bulbs). During that time, widespread advances in lighting and control 126 The NTSB investigated a September 25, 2001, incident in which a Boeing 757 took off from a closed runway at Denver International Airport, Denver, Colorado. A system failure affected the availability of a NOTAM about the runway closure, and a controller cleared the airplane to take off from the closed runway. After that incident, the flight crewmembers stated that they were unaware that the runway was closed, and the captain did not recall if the runway closure information was included in the ATIS broadcast. As a result of the incident, the NTSB recommended that the FAA “require the use of physical devices or other means to clearly indicate to flight crews of arriving and departing aircraft that a runway is closed” (A-03-5). In response, the FAA issued AC 150/5370-2F, “Operational Safety on Airports During Construction,” which indicated that airports should use physical devices or other means to indicate to flight crews that a runway is closed.

[ANALYSIS] Specifically, when pilots review information, their scan and retention of that information is influenced by the pilots’ perceived relevance of the information to a task. Thus, it is possible for a pilot to miss more relevant information when it is presented with information that is less relevant. Although human limitations (such as fatigue and time pressure/workload) may affect the review of information, these limitations can be overcome with effective information presentation. For example, items in the middle of a list can be better retained if the information is presented with “intensity” because a sharp, clear, or salient presentation has a better chance of being recalled compared with less visually stimulating information (FAA 2008). Therefore, the NTSB recommends that the FAA (1) establish a group of human factors experts to review existing methods for presenting flight operations information to pilots, including flight releases and general aviation flight planning services (preflight) and ACARS messages and other in-flight information; (2) create and publish guidance on best practices to organize, prioritize, and present this information in a manner that optimizes pilot review and retention of relevant information; and (3) work with air carriers and service providers to implement solutions that are aligned with the guidance. The NTSB notes that one way to ensure that air carriers effectively present and prioritize relevant information in flight releases and ACARS messages would be to develop an industry standard. 2.3.2 Role of Expectation Bias 2.3.2.1 Initial Taxiway Misalignment Because the flight crewmembers were not aware of the runway 28L closure, they were likely expecting SFO to be in its usual configuration, which would include runway 28L being open for departures and arrivals and airplanes using taxiway F (as shown in figure 3) to reach the departure end of runway 28L. Because of the runway 28L closure on the night of the incident, airplanes were using taxiway C to depart from runway 28R. The flight crewmembers stated, during postincident interviews, that the taxiway C surface resembled a runway, which they believed was runway 28R. A cue supporting the crewmembers’ perception that they were aligned with runway 28R was the lighting from the airplanes on taxiway C. Specifically, the airplanes’ wingtip navigation lights would have partially resembled (width-wise) runway edge lighting.100 Also, the airplanes’ flashing red beacon lights would have 99 The NTSB conducted an ASRS search to identify reports since 2010 related to flight crew confusion resulting from the depiction of ACARS messages and/or predeparture flight plan information. The NTSB found multiple reports of altitude and lateral deviations in which the pilot cited the formatting of information as a related issue, which further indicates that information display improvements are needed for text-based resources.

[ANALYSIS] The NTSB classified Safety Recommendation A-03-5 “Closed—Acceptable Action” on January 30, 2012. 127 The NTSB has investigated other events in which pertinent NOTAM information was missed; for example, see NTSB accident numbers DFW07CA092, CEN12LA229, and CEN14FA505. NTSB Aircraft Incident Report 65 technology have occurred. These advances could allow more conspicuous attentional capture features to be incorporated into a runway closure marker’s design. Such features, which include varying the flash pattern, incorporating strobe lights, and/or creating apparent movement, might direct a pilot’s attention to a closed runway better than the current design of the lighted “X.” The NTSB concludes that increased conspicuity of runway closure markers, especially those used in parallel runway configurations, could help prevent runway misidentification by flight crews while on approach to an airport. Therefore, the NTSB recommends that the FAA (1) conduct human factors research to determine how to make a closed runway more conspicuous to pilots when at least one parallel runway remains in use and (2) implement a method to more effectively signal a runway closure to pilots during ground and flight operations at night. NTSB Aircraft Incident Report 66 3. Conclusions 3.1 Findings 1. None of the following were factors in this incident: (1) flight crew qualifications, which were in accordance with Canadian and US regulations; (2) flight crew medical conditions; (3) airplane mechanical conditions; and (4) airport lighting, which met US regulations. 2. The first officer did not comply with Air Canada’s procedures to tune the instrument landing system (ILS) frequency for the visual approach, and the captain did not comply with company procedures to verify the ILS frequency and identifier for the approach, so the crewmembers could not take advantage of the ILS’s lateral guidance capability to help ensure proper surface alignment. 3. The flight crew’s failure to manually tune the instrument landing system (ILS) frequency for the approach occurred because (1) the Flight Management System Bridge visual approach was the only approach in Air Canada’s Airbus A320 database that required manual tuning of a navigation frequency, so the manual tuning of the ILS frequency was not a usual procedure for the crew, and (2) the instruction on the approach chart to manually tune the ILS frequency was not conspicuous during the crew’s review of the chart. 4. The first officer’s focus on tasks inside the cockpit after the airplane passed the final waypoint reduced his opportunity to effectively monitor the approach and recognize that the airplane was not aligned with the intended landing runway. 5. The flight crew-initiated, low-altitude go-around over the taxiway prevented a collision between the Air Canada airplane and one or more airplanes on the taxiway. 6. The controller responded appropriately once he became aware of the potential conflict. 7. Errors that the flight crewmembers made, including their false assumption that runway 28L was open, inadequate preparations for the approach, and delayed recognition that the airplane was not lined up with runway 28R, reflected breakdowns in crew resource management and led to minimal safety margins as the airplane overflew taxiway C. 8. The flight crewmembers’ lack of awareness about the runway 28L closure and the crewmembers’ previous experience seeing two parallel runways at San Francisco International Airport led to their expectation to identify two runway surfaces during the approach and resulted in their incorrect identification of taxiway C instead of runway 28R as the intended landing runway.

[ANALYSIS] Also, the captain stated that he had never seen runway 28L “dark” and that none of his previous landings at SFO occurred when a runway was closed. A runway closure marker with a flashing white “X” was placed at the threshold of runway 28L to indicate the runway closure. However, the flashing “X” would not have been in the flight crew’s direct field of view because the “X” was oriented toward the runway 28L final 95 Industry analysis of traffic patterns at SFO showed that airplanes landing on runway 28R tended to stay more to the right of course when runway 28L was open than when it was closed. NTSB Aircraft Incident Report 48 approach corridor and the airplane was not aligned with runway 28L.96 Also, the flash rate might have been too slow to capture the crew’s attention.97 (The conspicuity of the “X” to flight crews on final approach is further discussed in section 2.6.) The NTSB concludes that the flight crewmembers’ lack of awareness about the runway 28L closure and the crewmembers’ previous experience seeing two parallel runways at SFO led to their expectation to identify two runway surfaces during the approach and resulted in their incorrect identification of taxiway C instead of runway 28R as the intended landing runway. The NTSB considered the presentation and priority of the runway 28L closure information compared with other information that the flight crew received. The flight release package was 27 pages long and consisted of, among other items, route, weather, and NOTAM information. The NOTAM indicating the runway 28L closure (“RWY 10R/28L CLSD”) appeared on page 8 of the package, which was also the second page of NOTAM information, under the gray highlighted heading “DESTINATION” (which appeared on the previous page). Features of the NOTAM text emphasized the closure information, such as the use of bold font for the words “RWY” and “CLSD” and a “**NEW**” designation in red font with asterisks before the NOTAM text, as shown in figure 10.98 However, this level of emphasis was not effective in prompting the flight crewmembers to review and/or retain this information, especially given the NOTAM’s location (toward the middle of the release), which was not optimal for information recall. A phenomenon known as “serial position effect” describes the tendency to recall the first and last items in a series better than the middle items (Colman 2006). The ACARS message providing ATIS information Quebec, as displayed in the cockpit, was 14 continuous lines with all text capitalized in the same font. As shown in figure 11, the NOTAM indicating the runway 28L closure appeared at the end of line 8 and the beginning of line 9. The uniform presentation of the ATIS information could have contributed to the flight crew’s oversight of the runway closure information.

[ANALYSIS] The NTSB concludes that, if an airplane were to align with a taxiway, an automated ASDE alert would assist controllers in identifying and preventing a potential taxiway landing as well as a potential collision with aircraft, vehicles, or objects that are positioned along taxiways. The taxiway arrival prediction capability that was implemented at SEA (and was scheduled to be evaluated and implemented at other feasible ASDE-X locations) could be expanded to other ASDE system models (ASDE-3 and ASSC). Therefore, the NTSB recommends that the FAA modify ASDE systems (ASDE-3, ASDE-X, and ASSC) at those locations where the system could detect potential taxiway landings and provide alerts to air traffic controllers about potential collision risks. 2.6 Runway Closure Markers A runway closure marker with a lighted flashing white “X” appeared at the approach and departure ends of runway 28L when it was closed, including on the night of the incident. The lighted “X” was consistent with the specifications in FAA AC 150/5345-55A, “Specification for L-893, Lighted Visual Aid to Indicate Temporary Runway Closure.” Although the runway closure marker might have been effective at preventing a takeoff from or a landing on runway 28L when it was closed (the specific risks that the lighted “X” was designed to address), the runway closure marker did not capture the attention of the incident flight crew as the airplane approached the airport while aligned with taxiway C. The lighted “X” runway closure marker was not designed to address the possibility that a flight crew could misidentify a runway surface due to ineffective signaling of a runway closure. Although air traffic controllers can provide instructions to pilots about the closure of a runway, NOTAMs and ATIS broadcasts/messages are the primary means to inform pilots about runway closures. However, the information about runway closures provided in NOTAMs and ATIS broadcasts/messages is not necessarily a reliable means for ensuring that pilots are aware of the NTSB Aircraft Incident Report 64 closure information.126 As previously stated, although the runway 28L closure on the night of the incident was indicated in NOTAM and ATIS information that the incident flight crew received, that information was not effective in preventing the flight crew from misaligning the airplane during the approach to runway 28R.127 The incident flight crew and the flight crew of DAL521 (which landed on runway 28R about 4 minutes before the incident occurred) stated that they did not see a lighted “X” on runway 28L to indicate that it was closed. Given the current specifications for the runway closure marker, these flight crews likely did not see the “X” because their airplanes were not aligned with runway 28L. This situation highlights the need for a surface-based system with conspicuous and unambiguous visual cues that clearly indicate when a runway is closed, even if an approaching airplane is not aligned with the closed runway.

[ANALYSIS] Weather data obtained after the accident depicted the weather conditions in the area shortly before the time of the accident. For example, airport weather observation data showed that heavy rain—defined by the NWS as 0.03 inch of rain within 6 minutes—had begun falling at the airport by about 2338, 12 minutes before the accident. In addition, the National Lightning Detection Network detected 903 cloud-to-ground lightning strikes within 20 miles of the airport in the 15 minutes before the accident and 46 strikes within 5 miles of the airport in the 5 minutes before the accident; most of the strikes were located northeast, west, and southwest of the airport and along a line that was parallel to (and to the west of) flight 1420’s final approach path. American’s policy did not prohibit flight crews from continuing an approach with thunderstorms in the terminal area as long as the crews ensured that their intended route was clear of the thunderstorms. During the descent into the terminal area, there was no evidence to indicate that the route was not clear. In public hearing testimony, the first officer stated that, during the descent, the weather was to the left and moving off to the right and that the airport looked clear. Thus, the Safety Board concludes that, during the Analysis 115 Aircraft Accident Report descent into the terminal area, the flight crewmembers could have reasonably believed that they could reach the airport before the thunderstorm. Figure 18. Weather and Flight Information for 2339:12 2.2.1.2 Maneuvering to the Airport for Final Approach The flight crewmembers had been previously told by the controller to expect an instrument landing system (ILS) approach to runway 22L. At 2339:45, the local controller broadcast the first of two windshear alerts, reporting that the centerfield wind was 340º at 10 knots.178 At this point, the airplane was about 8 miles from the airport. Afterward, the flight crew requested a change in runways from 22L to 4R because the winds had shifted to the northwest. The use of runway 22L could have resulted in a tailwind at the time of landing, so runway 4R was a more appropriate choice because of the expectation of a headwind. (NWS radar data indicated that the leading edge of a line of thunderstorms was over the airport at this time with the heaviest activity located northwest through northeast 5 miles from the runway.)179 Figure 19 shows a map of the weather conditions at 2339:45 and flight 1420’s location. 178 The windshear alert also indicated that the north boundary wind was 330º at 25 knots and that the northwest boundary wind was 010º at 15 knots. 179 The leading edge is one of the most hazardous areas in thunderstorms because of the updraft-downdraft interaction. Analysis 116 Aircraft Accident Report Figure 19.

[ANALYSIS] Weather radar data indicated that heavy rain was occurring at the airport at the time that the airplane was flying on the 220° heading.) Figure 21 shows a map of the weather conditions at 2344:30 and flight 1420’s location. Figure 21. Weather and Flight Information for 2344:30 181 At 2343:26, the first officer said, “…there’s the airport right there. Okay?” Five seconds later, the captain asked, “where?” At 2343:31, the first officer said, “okay, you’re set up on a base for it. Okay?” to which the captain questioned, “I’m on a base now?” At 2343:35, the first officer said, “well, you’re on a dog leg. You’re comin’ in. There’s the airport.” Three seconds later, the captain said, “I lost it,” and the first officer said, “right there, you’re you’re downwind. See it’s right there.” The captain then said, “I still don’t see it...well just vector me. I don’t know.” Analysis 119 Aircraft Accident Report The weather south of the airport became the next focus of the flight crew’s attention. NWS radar data indicated that, in the 6 minutes before the accident, a line of thunderstorms, with several large areas of intense and extreme activity, was encompassing the Little Rock airport area and runway 4R approach path. At 2345:47, the first officer told the controller “we’re getting pretty close to this storm. we’ll keep it tight if we have to.” The controller indicated to the flight crew that, “when you join the final, you’re going to be right at just a little bit outside the marker if that’s gonna be okay for ya.” The captain stated, “that’s great,” and the first officer told the controller, “that’s great with us.” Figure 22 shows a map of the weather conditions at 2345:47 and flight 1420’s location. Figure 22. Weather and Flight Information for 2345:47 The airplane was once again turned westbound, and the controller provided the flight crew with heading instructions to position the airplane to intercept the runway 4R final approach course. At 2346:39, the controller informed the flight crew that the airplane was 3 miles from the outer marker, instructed the crew to maintain 2,300 feet until the airplane was established on the localizer, and cleared the crew for the runway 4R ILS approach. In a postaccident interview, the first officer stated that, at this point, there was urgency to land because the weather was “up against” the airport. However, the first officer also stated that, after being vectored to the runway 4R ILS approach course, he had visual contact with the runway throughout the rest of the approach. At 2346:52, the controller advised the flight crew that heavy rain was falling at the airport, visibility was less than 1 mile, ATIS information Romeo was no longer current, and the runway visual range (RVR) was 3,000 feet.

[ANALYSIS] 2. Analysis 2.1 General The captain and the first officer of American Airlines flight 1420 were properly certificated and qualified under Federal and company requirements. No evidence indicated any preexisting medical or behavioral conditions that might have adversely affected the flight crew’s performance during the accident flight. The accident airplane was properly certified, equipped, and maintained in accordance with Federal regulations and approved company procedures. No evidence indicated preexisting engine, system, or structural failures. This analysis focuses primarily on the flight crew’s performance and the airplane’s spoiler system. The flight crew’s performance on approach to the airport is examined during three segments of the approach and in the context of the weather information and cues that were available. The flight crew’s and the airplane’s performance during the landing and overrun sequences are also examined. The analysis also addresses the roles of situational stress and fatigue in the accident sequence; meteorological support, including ATC services; emergency response efforts; airport issues; and American and FAA oversight. 2.2 Accident Scenario 2.2.1 The Approach The flight crew had multiple sources of information indicating that thunderstorms might become a factor during the approach. The preflight weather package (discussed in section 1.7.3) included two NWS weather advisories for severe thunderstorms and a company SIGMEC [significant meteorological condition] for a widely scattered area of thunderstorms along the planned route. Also, the dispatcher’s 2254 message via the aircraft communication addressing and reporting system informed the crew of the lines of thunderstorms to the left and right of the planned route and suggested that the crew expedite the arrival to the airport “to beat the thunderstorms.” In addition, NWS Convective SIGMET [significant meteorological information] 15C, received about 2304, warned of a line of severe thunderstorms moving southeast through Arkansas with hail up to 2 inches in diameter and the possibility of wind gusts to 70 knots. Finally, automatic terminal information service (ATIS) information Romeo, which was current beginning about 2326, indicated that a thunderstorm with frequent lightning was located west through northwest of the airport and moving northeast. Analysis 114 Aircraft Accident Report The flight crewmembers also had weather information from the airborne weather radar and their view outside the cockpit. Statements from the CVR indicated that the flight crew had discussed the weather and the need to expedite the approach.

[ANALYSIS] As with runway 4R/22L in Little Rock, runway 8/26 in Burbank was exempt, under 14 CFR 139.309(a)(1), from the 1,000-foot runway safety area standard in AC 150/5300-13. On December 13, 1994, the Safety Board issued Safety Recommendation A-94-211, which asked the FAA, among other things, to require that substandard runway safety areas be upgraded to AC 150/5300-13 minimum standards wherever possible.229 In its October 15, 1997, response, the FAA indicated that 25 percent of the runways at 14 CFR Part 139 certificated airports have safety areas that do not meet AC 150/5300-13 minimum standards but could with feasible improvements and that 17 percent have safety areas that could not be feasibly improved to meet the standard. However, the FAA stated that runway safety area improvement projects would be scheduled only as part of overall 227 Construction of runway 4R/22L began as early as September 1982. Work on the runway continued after January 1, 1988, but none of the efforts involved reconstruction or significant expansion. 228 The description for this accident, DCA00MA030, can be found on the Safety Board’s Web site at <http://www.ntsb.gov>. 229 This recommendation was issued as a result of the April 27, 1994, Action Air Charters flight 990 accident in Stratford, Connecticut (see section 1.18.8.1). Analysis 159 Aircraft Accident Report runway improvement projects because of the associated cost and infrequency of aircraft overruns and undershoots. On February 10, 1999, the Board expressed its concern that the delay in runway safety area upgrades would allow nonstandard conditions to continue and classified Safety Recommendation A-94-211 “Closed—Unacceptable Action.” The Safety Board recognizes that the design of some airport runways makes it difficult for runway safety areas to be upgraded to the standards of AC 150/5300-13 and that those airport runways that can be upgraded may not be improved for some time based on the FAA’s current plans. However, those runways should provide equivalent runway protection. One way to achieve this goal is to install a type of soft-ground aircraft arresting system, such as the Engineering Materials Arresting System (EMAS). The safety benefit of EMAS was demonstrated by the American Eagle flight 4925 accident when the airplane departed an 8,400-foot runway but was stopped approximately 248 feet into a 400-foot EMAS (see section 1.16.5). According to a report by Engineered Arresting Systems Corporation, which developed EMAS, the flight 1420 airplane would not have been significantly slowed by a standard EMAS installed at the approach end of runway 22L because the airplane’s track was outside the extended runway edges.

[ANALYSIS] However, the first officer also stated that, after being vectored to the runway 4R ILS approach course, he had visual contact with the runway throughout the rest of the approach. At 2346:52, the controller advised the flight crew that heavy rain was falling at the airport, visibility was less than 1 mile, ATIS information Romeo was no longer current, and the runway visual range (RVR) was 3,000 feet. Simultaneously, the captain Analysis 120 Aircraft Accident Report stated, “aw, we’re goin’ right into this.” Figure 23 shows a map of the weather conditions at 2346:52 and flight 1420’s location. About this point in the approach, the flight crew should have regained the use of the airborne weather radar to depict precipitation levels in front of the airplane. However, the Safety Board could not determine whether the radar was displaying this information or whether the flight crewmembers, given their workload, were able to perceive and interpret the information that the radar was providing. Figure 23. Weather and Flight Information for 2346:52 At 2347:08, the controller cleared flight 1420 to land and stated that the wind was 350º at 30 knots gusting to 45 knots. The CVR indicated that flight crew did not discuss the wind information, the heavy rain that was already falling at the airport, or the depiction of the weather on the airborne weather radar. Figure 24 shows a map of the weather conditions at 2347:08 and flight 1420’s location. Statements from the CVR and the first officer’s postaccident interviews indicated the flight crew’s concern about the location of the thunderstorm in relation to the airport and the airplane. However, the Safety Board concludes that, because the first officer was able to maintain visual contact with the runway as the airplane was vectored for the final approach course, both flight crewmembers might still have believed that flight 1420 could arrive at the airport before the thunderstorm. The Board understands that some other flight crews might continue an approach to a runway under the same circumstances. On the other hand, the Board also recognizes that the approaching storm and the reports of heavy rain, dropping visibility, and increasing crosswinds (from 10 to 30 knots with Analysis 121 Aircraft Accident Report gusts to 45 knots) would be sufficient for some flight crews to hold until the storm passed or proceed to an alternate airport. Figure 24. Weather and Flight Information for 2347:08 2.2.1.3 Final Approach Segment The final approach segment began when the airplane intercepted the localizer, at 2347:16. Simultaneously, the first officer erroneously read back the controller’s previous wind report of 350º at 30 knots gusting to 45 knots as “zero three zero at four five.” The flight crew then discussed the effect of the previously reported RVR on the approach.

[CONCLUSIONS] Automatic brake systems reduce pilot workload during landings in wet, slippery, or high crosswind conditions. 16. The lack of spoiler deployment was the single most important factor in the flight crew’s inability to stop the accident airplane within the available runway length. 17. The flight crewmembers’ performance during the accident flight was degraded, as evidenced by their operational errors and impaired decision-making. 18. The flight crewmembers’ focus on expediting the landing because of the impending weather contributed to their degraded performance. 19. Aircraft penetration of thunderstorms occurs industry-wide. 20. The flight crew’s degraded performance was consistent with known effects of fatigue. 21. The local controller provided appropriate, pertinent, and timely weather information to the flight crew regarding the conditions on approach to and at the airport. 22. If near-real-time color weather radar showing precipitation intensity were available, it would provide air traffic controllers with improved representation of weather conditions in their areas of responsibility. 23. The ability of flight dispatchers to provide timely and accurate weather support would be enhanced if they had access to Terminal Doppler Weather Radar information at airports where it is available and Weather Systems Processor information when the system becomes available. 24. Center Weather Service Units should be staffed at all times when any significant weather is predicted to affect their areas of operation, even if the weather is predicted to occur before or after normal operating hours. Conclusions 169 Aircraft Accident Report 25. The Automated Surface Observing System lockout period can prevent the relay of critical weather information to flight crews. 26. Runway visual range data should be directly reported to automated weather systems. 27. The current 2-hour runway visual range archiving capability is inadequate to ensure that data can be preserved for future use. 28. If detailed information on the Low Level Windshear Alert System were contained in the Federal Aviation Administration’s Aeronautical Information Manual, pilots could have a better understanding of the system. 29. Part of the delay in locating the flight 1420 wreckage was preventable, and several minutes in the emergency response time might have been saved if the Aircraft Rescue and Fire Fighting units had proceeded directly to the departure end of runway 4R. 30. Aircraft Rescue and Fire Fighting (ARFF) units may not be staffed at a level that enables ARFF personnel, upon arrival at an accident scene, to conduct exterior firefighting activities, an interior fire suppression attack, and a rescue mission. 31. A crash detection and location technology would help expedite the arrival of emergency responders at an accident scene, thus maximizing the possibility for saving lives and reducing the severity of injuries. 32. A timely postaccident interagency emergency response critique that identifies deficiencies that need corrective action and successes that should be repeated in similar circumstances would be beneficial for all parties involved in an aviation accident response. 33. The development of recent technologies to convert nonfrangible structures to frangible ones would provide a safety benefit to airport facilities. 34. American Airlines has insufficient guidance to assist its pilots in performing a stabilized approach and recognizing when an approach has become unstabilized. 35.

[ANALYSIS] Thus, the flight 1420 airplane would not have been able to use the full length of a standard EMAS. However, according to FAA AC 150/5220-22, “Engineered Materials Arresting System for Aircraft Overruns,” most airplane runway overruns “come to rest within 1,000 feet of the runway end and between the extended edges of the runway.” Therefore, the Safety Board continues to support the installation of EMAS, especially for those runways in which the safety area is less than the minimum standards established in AC 150/5300-13. The Board notes that an EMAS was installed at the departure end of runway 4R at Little Rock in the fall of 2000. The Board further notes that Little Rock airport is working with Federal and local government agencies to extend the runway safety area at the departure end of runway 4R to 1,000 feet by July 2002. 2.5.2 Nonfrangible Structures The FAA determined that the runway 22L approach lighting system at Little Rock, which is located in a flood plain area of the Arkansas River, could not be retrofitted to a frangible design because of the possibility that moving water, ice, and floating debris would affect the structural integrity of the system. The Safety Board recognizes the current design limitations of this approach lighting system and acknowledges that, if the approach lighting system had been frangible, it is possible that the accident airplane would not have been stopped on the ground and would have gone into the Arkansas River. However, the Board also recognizes that frangible structures, because of their ability to break, distort, or yield on impact with aircraft, generally present less risk than nonfrangible ones. In this accident, the airplane’s collision with the nonfrangible approach lighting system was the direct cause of the fatal blunt force trauma injuries sustained by the captain and the passengers in seats 3A and 8A and the destruction to the airplane on the left side of the fuselage. Analysis 160 Aircraft Accident Report In 1984, the Safety Board issued Safety Recommendation A-84-36, which asked the FAA to “initiate research and development activities to establish the feasibility of submerged low-impact resistance support structures for airport facilities and promulgate a design standard if such structures are found to be practical.” The FAA conducted research in this area with the National Institute of Standards and Technology. In October 1996, the FAA concluded that, with the current technology, any submerged low-impact frangible structure would most likely be destroyed by wave motion from small storms.

[ANALYSIS] Finally, automatic terminal information service (ATIS) information Romeo, which was current beginning about 2326, indicated that a thunderstorm with frequent lightning was located west through northwest of the airport and moving northeast. Analysis 114 Aircraft Accident Report The flight crewmembers also had weather information from the airborne weather radar and their view outside the cockpit. Statements from the CVR indicated that the flight crew had discussed the weather and the need to expedite the approach. For example, at 2325:47, the captain said, “we got to get over there quick.” The first officer then stated, “I don’t like that...that’s lightning,” to which the captain replied, “sure is.” At 2328:30, the captain repeated, “we gotta get there quick.” At 2329:55, the first officer stated, “I say we get down as soon as we can.” The CVR also indicated that the crew had the city of Little Rock and the airport area in sight by at 2326:59. Further, the CVR recorded the captain’s announcement to the passengers, beginning at 2327:31, that “quite a light show” was to the left of their course and that they would be passing the lightning on the way to Little Rock. 2.2.1.1 Descent Into the Terminal Area At 2334:11, the local controller told the flight crew, upon initial contact, that a thunderstorm located northwest of the airport was “moving through the area now” and that the winds were from 280º at 28 knots gusting to 44 knots. Even though the flight crewmembers had previously discussed the need to expedite the approach because of the weather, the CVR indicated that they had not discussed the possibility that the thunderstorm might reach the airport before the flight landed. In a postaccident interview, the first officer stated that, during the descent, the weather appeared to be about 15 miles away from the airport and that he and the captain thought that there was “some time” to make the approach. At 2339:12, the first officer told the controller, “that storm is moving this way like your radar says it is but a little bit farther off than you thought.” At that point, flight 1420 was about 11 miles south of the airport. Figure 18 shows a map of the weather conditions at 2339:12 and flight 1420’s location. After receiving the controller’s next wind report (330º at 11 knots), the first officer indicated that the winds were “a little bit better” than they had been earlier. Weather data obtained after the accident depicted the weather conditions in the area shortly before the time of the accident. For example, airport weather observation data showed that heavy rain—defined by the NWS as 0.03 inch of rain within 6 minutes—had begun falling at the airport by about 2338, 12 minutes before the accident.

[ANALYSIS] Aircraft Accident Report NTSB/AAR/89-04. Washington, DC. Analysis 158 Aircraft Accident Report 2.5 Airport Factors 2.5.1 Runway Safety Areas Runway 4R/22L, which was opened in September 1991, has runway safety areas of 1,000 feet at the departure end of 22L and 450 feet at the departure end of 4R. Although the FAA’s June 5, 1991, version of AC 150/5300-13 stated that the standard runway safety area was 1,000 feet, runway 4R/22L was exempt from this standard under the provisions of 14 CFR 139.309(a)(1), which allowed runways that had a safety area on December 31, 1987, to be maintained, as long as no reconstruction or significant expansion of the runway had begun after January 1, 1988.227 The Safety Board notes that, in this accident, an extra 550 feet at the departure end of runway 4R would not have prevented the airplane from departing the end of the runway or impacting the approach lighting system; however, the airplane’s speed would have further decreased with an extra 550 feet at the end of the runway, resulting in a lower impact speed. Because safety areas of at least 1,000 feet would provide an extra margin of safety under most circumstances, the Board is concerned about runway safety areas that are less than the current FAA standard. Another recent accident involved an overrun beyond the threshold of a runway with a nonstandard safety area. Specifically, on March 5, 2000, Southwest Airlines flight 1455, a Boeing 737-300, N668SW, departed the end of runway 8 during landing at Burbank-Glendale-Pasadena Airport, Burbank, California. The airplane traveled through a nonfrangible metal blast fence beyond the departure end of the runway and came to rest on a highway outside the airport perimeter. Of the 142 airplane occupants, 2 received serious injuries, 42 received minor injuries, and 98 were uninjured.228 The runway safety area at the approach (west) end of runway 8/26 is 200 feet; no safety area exists at the departure (east) end of the runway. As with runway 4R/22L in Little Rock, runway 8/26 in Burbank was exempt, under 14 CFR 139.309(a)(1), from the 1,000-foot runway safety area standard in AC 150/5300-13.

[ANALYSIS] The ARFF units proceeded to the approach end of runway 4R, but the airplane was off the departure end of the runway. As a result, the ARFF units had to travel back to the taxiway at which they entered the runway and then proceed to the other end of the runway. The ARFF units located the airplane about 0003, 11 minutes after the initial call from the local controller. However, they did not arrive on scene until 5 minutes later, about 0008 (16 minutes after the initial notification), because they had to travel in the opposite direction to an access road, turn onto a perimeter road back in the direction of the accident site, stop to manually unlock a perimeter security gate, and then continue on the perimeter road to the accident site. If the ARFF units had known the approximate location of the airplane when they left the fire station, the time spent traveling from the taxiway to the approach end of the runway and back would have been saved. The ARFF units reported that they were initially traveling very slowly because of the limited visibility toward the approach end of the runway and the unknown location of the airplane. The Safety Board recognizes that the controller could have provided a more precise description of the accident location to the ARFF units, especially since he knew the direction in which the airplane was landing and had seen the airplane travel past midfield. However, the Board also recognizes that the ARFF personnel could have queried the controller to see if he knew any additional information about the airplane’s location. Analysis 153 Aircraft Accident Report The Safety Board concludes that part of the delay in locating the flight 1420 wreckage was preventable and that several minutes in the emergency response time might have been saved if the ARFF units had proceeded directly to the departure end of runway 4R. Because the delay can be partly attributed to the incomplete location information provided to the ARFF units by the local controller, the Safety Board believes that the FAA should issue a mandatory briefing item to tower controllers that describes the circumstances of this accident, including the interactions between the controller and ARFF crews. This briefing item should emphasize that location information provided to ARFF crews should be as complete and specific as possible to minimize opportunities for confusion. The Safety Board also believes that the FAA should amend FAA Order 7110.65, “Air Traffic Control,” to require controllers to monitor the progress of ARFF crews responding to emergencies to ensure that the response is consistent with known location information. In addition, the Safety Board believes that the FAA should amend FAA Order 7210.3R, “Facility Operation and Administration,” to direct tower managers to establish mutual annual briefings between ATC and ARFF personnel to ensure that these personnel have a common understanding of the local airport emergency plan and sections of the FAA’s AC 150/5210-7C, “Aircraft Rescue and Firefighting Communications,” that are applicable to local ATC/ARFF emergency response procedures.

[ANALYSIS] Operations on contaminated runways  MD-88 pilots received a flight safety electronic bulletin discussing contaminated runway awareness.  Delta was sponsoring two demonstration studies of aircraft-based technology for conducting runway friction assessments on contaminated runways. Special winter operations airport program  Delta identified LGA as a special winter operations airport for the 2015-2016 winter season. (As previously stated, LGA was not identified as a special winter operations airport for the 2014-2015 winter season, during which time the accident occurred.) NTSB Aircraft Accident Report 74  A Delta senior vice president sent a letter to Port Authority “urging” the adoption of a more robust friction assessment program.  Delta made “repeated requests” to LGA to issue hourly field condition reports, and LGA agreed (as discussed in section 2.6.2).  In December 2015, Delta’s flight safety department initiated a review of the special winter operations airport program to assess its effectiveness. The group conducting the review determined that the program was effective in identifying and mitigating potential runway excursion risks during winter operations. The group identified changes to enhance the program’s assessment of airports and the methods to mitigate identified hazards. Also, the group recommended several modifications to address the upcoming changes resulting from the TALPA program. NTSB Aircraft Accident Report 75 3. Conclusions 3.1 Findings 1. The flight crew was properly certificated and qualified in accordance with federal regulations and company requirements. Flight crew fatigue was likely not a factor in this accident. The airplane was properly certificated, equipped, and maintained in accordance with federal regulations. No evidence indicated any preimpact structural, engine, or system failures. 2. The flight crewmembers’ uncertainty about the runway conditions at LaGuardia Airport led to some situational stress for the captain. 3. The flight crew was well prepared for the approach and established landing requirements that were consistent with company policies. 4. Even though the flight crewmembers’ observations of snow on the runway were inconsistent with the expectations that they formed based on the field condition information that they received, their decision to continue the approach was not inappropriate because the landing criteria had been met. 5. Although the runway was contaminated with snow, runway friction when the accident airplane landed was sufficient for stopping on the available runway length. 6. The circumstances associated with the landing, including the snowier-than-expected runway, short runway length, and body of water off the departure end of the runway, likely exacerbated the captain’s situational stress and prompted him to make an aggressive input on the thrust reversers. 7. The captain was unable to maintain directional control of the airplane due to rudder blanking, which resulted from his application of excessive reverse thrust. 8. Even though the first officer promptly identified rudder blanking as a concern and the captain stowed the thrust reversers in response, the airplane’s departure from the left side of the runway could not be avoided because directional control was regained too late to be effective. 9. A solution to reliably limit reverse thrust engine pressure ratio values could benefit all pilots of MD-80 series airplanes. 10.

[ANALYSIS] At 1104:00, the snow coordinator notified the local controller that runway 13 was closed and then proceeded onto the runway in his vehicle. Because the controller had not responded to the snow coordinator’s notification, he repeated that the runway was closed. At 1104:16, the local controller asked the snow coordinator whether the runway was closed, and the snow coordinator confirmed that information. About 14 seconds later (and 33 seconds after the initial notification that the runway was closed), the controller instructed the flight crew of Delta flight 1999 (the next arriving airplane for runway 13) to go around. On the basis of airport protocol and the letter of agreement with the ATC tower, the snow coordinator had likely assumed that the controller would have immediately closed the runway. However, this assumption led to a situation in which the snow coordinator’s vehicle was on the runway for about 23 seconds while Delta flight 1999 was on final approach. At the time of the controller’s go-around instruction, the flight 1999 airplane was only about 30 seconds from landing. The controller did not expect an abrupt closure of the runway and had likely assumed that the closure was due to the winter weather. (The snow coordinator did not state that the accident was the reason for the closure.) However, the controller had just attempted to contact the accident airplane’s flight crew six times without a response, so he might have reacted more quickly if he had integrated that information along with the runway closure information. Airport operations personnel notified the airport operations manager via cell phone that an airplane had departed the paved runway surface, hit a fence, and was leaking fuel. At the time, the airport operations manager happened to be engaged in a face-to-face conversation with the ARFF deputy chief and relayed the phone call information to him. At 1104:35, the deputy chief called the ARFF crew chief to prepare ARFF crews for a response based on the phone call information. However, because the airplane’s precise location was not provided as part of the NTSB Aircraft Accident Report 68 initial cell phone call, the only information that the ARFF crews had regarding the accident location was that the airplane had hit a fence. At 1104:38, a responding airport operations staff member notified the local controller about the accident. At 1104:48, another responding airport operations staff member told the controller, “you have an aircraft off [runway] 3-1 on the north vehicle service road. Please advise crash/rescue.” However, the EANS—the primary method for communicating an emergency at LGA—was not immediately activated by ATC tower personnel. The snow coordinator arrived on scene and told the controller at 1105:55 that the airplane was “leaking fuel on the left side of his aircraft heavily…his wing is ruptured.” Thirty seconds later (and about 3.5 minutes after the airplane came to a stop), ATC tower personnel activated the EANS, which reported “LaGuardia, Alert 3, all emergency vehicles respond.

[CONCLUSIONS > FINDINGS] A callout when reverse thrust exceeds 1.3 engine pressure ratio during landings on contaminated runways could help avoid rudder blanking and a subsequent loss of directional control. 11. An automated alert could help minimize the possibility of reverse thrust engine pressure ratio exceedances during the landing rollout. NTSB Aircraft Accident Report 76 12. This accident demonstrates the continuing need for objective, real-time, airplane-derived data about runway braking ability for flight crews preparing to land with runway surface conditions that are worse than bare and dry. 13. The flight and cabin crews did not conduct a timely or an effective evacuation because of the flight crew’s lack of assertiveness, prompt decision-making, and communication and the flight attendants’ failure to follow standard procedures once the captain commanded the evacuation. 14. The flight attendants were not adequately trained for an emergency or unusual event that involved a loss of communications after landing, and the flight attendants’ decision to leave their assigned exits unattended after the airplane came to a stop resulted in reduced readiness for an evacuation. 15. This and other accidents demonstrate the need for improved decision-making and performance by flight and cabin crews when faced with an unplanned evacuation. 16. Aircraft rescue and firefighting personnel would likely have arrived at the accident scene sooner if they had received more timely and precise information about the accident and its location. 17. Even though the initial uncertainty regarding the total number of passengers aboard the accident flight, including lap-held children, did not lead to any adverse outcomes, such uncertainty could be detrimental under other accident circumstances, especially if search and rescue efforts are needed. 18. By not using its continuous friction measuring equipment during winter operations, LaGuardia Airport did not take advantage of a tool that would allow the airport to objectively assess the effectiveness of snow removal operations on contaminated runways. 19. The Federal Aviation Administration’s airport winter operations safety guidance is not sufficiently clear about the timing and need for updated runway condition reports, which could result in flight crew uncertainty about possible runway contamination. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the captain’s inability to maintain directional control of the airplane due to his application of excessive reverse thrust, which degraded the effectiveness of the rudder in controlling the airplane’s heading. Contributing to the accident were the captain’s (1) situational stress resulting from his concern about stopping performance and (2) attentional limitations due to the high workload during the landing, which prevented him from immediately recognizing the use of excessive reverse thrust. NTSB Aircraft Accident Report 77

[ANALYSIS] However, it would have been difficult for the crew to visually assess the nature and depth of the snow on the runway.88 Also, little time was available for the flight crew to reevaluate the decision to continue. Only 13 seconds had elapsed between the time that the captain called the runway in sight and the time of the 50-foot automated callout, when the captain would have been preparing to flare the airplane. During these 13 seconds, the captain was engaged in several tasks, including adjusting the aim point to land closer to the approach end of the runway, decrabbing the airplane to align its longitudinal axis with the runway, and making aileron inputs to correct for drift. The NTSB concludes that, even though the flight crewmembers’ observations of snow on the runway were inconsistent with the expectations that they formed based on the field condition information that they received, their decision to continue the approach was not inappropriate because the landing criteria had been met. 2.1.5 The Landing The airplane touched down about 17 seconds after the captain called the runway in sight. During the 27 minutes between the time that runway 13 was last cleared of snow and the landing, snow continued to fall, with fresh snow covering the runway. As previously stated, the 0903 NOTAM, which was current at the time of the accident, reported 1/4 inch of wet snow on runway 13. The NOTAM was issued more than 1.5 hours before runway 13 was last cleared of snow (as discussed further in section 2.6.2); however, given the amount of time between the last runway clearing and the landing and the likely amount of snow that had accumulated during that time, the NOTAM likely described the approximate depth of the contamination on the runway surface at the time of the airplane’s landing. The accident airplane was the fifth arrival on runway 13 after it was last cleared. During this time, one report of medium braking action at the touchdown zone was received about 87 Although the flight crewmembers reported, after the accident, that the appearance of the runway was not what they expected based on the field condition information, the CVR showed that they did not consider that the snow that was falling could alter the landing performance calculations or that the braking conditions might have deteriorated from the good braking action reported by the flights crews of two airplanes that had recently landed on runway 13. However, the flight crew was not required to perform a go-around maneuver because the landing criteria were met and, as shown by FDR and radar data, the captain flew a stabilized approach. 88 Company landing criteria indicated that landing was not permitted with a snow depth of 1 inch or greater. Given the recent reports of good braking action on runway 13, the flight crew had no information that would have specifically prohibited the landing. NTSB Aircraft Accident Report 52 19 minutes before the accident landing, and reports of good braking action were received about 16 and 8 minutes before the accident landing.89 Because braking action reports are inherently subjective, the NTSB considered the results of the simulations it conducted with Boeing as part of this accident investigation.

[ANALYSIS] The captain decided to continue the approach to a landing because he and the first officer had determined that the landing criteria had been met, including the landing distance performance requirements for good braking action. The captain called the runway in sight as the airplane descended through an altitude of about 233 feet agl. Besides company and ATIS reports indicating that the runway was plowed, sanded, and chemically treated and was wet with snow, the flight crew had overheard ATC stating, as late as 1040, that arriving airplanes were being held for runway cleanup. The first officer stated that these communications “painted a picture” of what he and the captain could expect to see once the airplane broke out of the clouds—a runway that had at least some patches of runway surface visible. However, the captain and the first officer reported that the runway appeared white rather than dark or patchy, which was not 85 ATIS information Quebec (which was current at the time) reported the visibility as 1/4 mile, which was below the 1/2-mile visibility required for the approach. However, the approach could still be performed because the RVR was above 2,400 feet. 86 Before final approach, the flight crew had not discussed the possible effects that a tailwind could have on landing performance. One explanation for that could be that, at the time that the captain conducted the approach briefing, the reported wind (from ATIS information Papa) was from 040º at 7 knots, resulting in a direct crosswind (90º) for runway 13 with no tailwind component. Another explanation could be that Delta’s training program did not include instruction on landing with a tailwind and that the company’s MD-88 adverse weather operating procedures did not provide guidance for landing with a tailwind on a contaminated runway. NTSB Aircraft Accident Report 51 consistent with their expectations regarding the nature and extent of the contamination on the runway given the recent snow cleaning operations and the reports of good braking action.87 Delta’s MD-88/90 operating procedures for adverse weather indicated that, “when there is contamination on the runway…captains must evaluate crew, aircraft, and environmental conditions in determining the safety of operating their flight.” Key environmental conditions included braking action reports, the nature of contamination (water, wet snow, or dry snow), and the depth of the contamination. The flight crew knew that two preceding airplanes had reported good braking action on the runway. However, it would have been difficult for the crew to visually assess the nature and depth of the snow on the runway.88 Also, little time was available for the flight crew to reevaluate the decision to continue. Only 13 seconds had elapsed between the time that the captain called the runway in sight and the time of the 50-foot automated callout, when the captain would have been preparing to flare the airplane.

[ANALYSIS] NTSB Aircraft Accident Report 47 2. Analysis 2.1 Accident Sequence 2.1.1 General The flight crew was properly certificated and qualified in accordance with federal regulations and company requirements. Flight crew fatigue was likely not a factor in this accident.77 The airplane was properly certificated, equipped, and maintained in accordance with federal regulations. No evidence indicated any preimpact structural, engine, or system failures. 2.1.2 Preflight Activities A major winter storm was impacting the northeastern United States on the day of the accident, with a significant accumulation of snow expected in the New York area. The flight crewmembers were aware of the storm, and they began preparing for the flight to LGA after arriving in ATL at 0705. While at the departure gate, the crewmembers reviewed information regarding the expected conditions at the time of landing, including Delta’s terminal forecast for the planned arrival time (1055), which indicated that visibility would be 1/2 mile in moderate snow and mist with a broken ceiling at 700 feet agl and 4 to 7 inches of snow expected. The forecasted conditions were expected through 1200 that day.78 The flight crew also reviewed a NOTAM issued at 0445 that indicated that the runways were wet and had been sanded and deiced with solid chemical. (A more recent NOTAM, issued at 0738, was not part of the 0740 company weather package provided to the flight crew; that NOTAM reported that runway 13 was covered with thin, wet snow.) In addition, the flight crew reviewed the ATIS information issued at 0751 (Lima) and 0851 (Mike).79 ATIS information Lima and Mike both reported that 77 The captain reported needing 8 to 9 hours of sleep per night to feel rested, so he had a cumulative sleep deficit of 3 to 3.5 hours (see table 2) during the 72 hours preceding the accident. The first officer reported needing 7.5 hours of sleep per night to feel rested, so he had a cumulative sleep deficit of about 2 hours (see table 3) during that period.

[ANALYSIS] However, the NTSB’s aircraft performance study showed that medium braking action conditions were present when the previous Delta MD-88 successfully stopped on runway 13. NTSB Aircraft Accident Report 49 to the first officer that company dispatch had not alerted them about the possibility of holding. The captain then asked the first officer if he had received any braking action reports from the controller. After the first officer replied that he had not yet received any reports, the captain repeated his frustration about dispatch’s lack of advanced notice about holding and expressed further frustration about the lack of timely braking action reports from ATC and dispatch (before the runway was closed for snow removal operations). The NTSB concludes that the flight crewmembers’ uncertainty about the runway conditions at LGA led to some situational stress for the captain. At 1045:38, the accident flight crewmembers heard a controller at the New York TRACON advise a flight crew of a poor braking action report for runway 13. The captain told the first officer “we can’t land with poor.” Shortly afterward, the controller advised another flight crew that an Airbus airplane had reported good braking action on runway 13. Afterward, the flight crew of a regional jet also reported that the braking action on runway 13 was good. During postaccident interviews, the accident captain recalled thinking that two reports indicating good braking action meant that conditions were adequate to proceed with the approach, and the accident first officer recalled thinking that two consecutive reports of good braking action were a “green light” to continue. ATIS information Quebec (issued at 1051) indicated that 1/4 inch of wet snow was observed on the runways at 0904 and that “all runways are wet and have been sanded and deiced with solid chemical.”82 However, the most recent NOTAMs at the time (issued at 0902 and 0903) reported that 1/4 inch of wet snow was on the runways but did not indicate that the runways had been treated. Further, according to radio communications between ATC and the LGA snow coordinator, about 1038 the snow coordinator reported to the local controller that “the runways have not been treated” and that “we’re just brooming and plowing.”83 (The flight crew did not know the information reported by the snow coordinator.) Thus, the ATIS report that was current at the time of the accident (Quebec), as well as the four previous ATIS reports (Lima, Mike, Oscar, and Papa, issued between 0751 and 1024), contained outdated and contradictory field condition information about the status of LGA’s runways. Because braking action reports are subjective, the captain attempted to assess the reliability of the reports indicating good braking action.84 At 1055:34, he asked the first officer, “wonder who reported braking action good?

[ANALYSIS] NTSB Aircraft Accident Report 60 runway.105 Postaccident simulations indicated that medium or greater braking action conditions were present at the time that the preceding airplane and the accident airplane touched down. (The flight crew of the preceding airplane did not provide a braking action report after the airplane landed.) Although the RCAM will provide operators with a framework for more objective runway surface condition reports, the use of airplane-based systems to determine runway surface conditions would be the most objective and timely method for reporting runway conditions to the pilots of arriving aircraft, especially since precipitation, temperature, and traffic can result in continuous changes to runway surface conditions. The NTSB concludes that this accident demonstrates the continuing need for objective, real-time, airplane-derived data about runway braking ability for flight crews preparing to land with runway surface conditions that are worse than bare and dry. The NTSB issued Safety Recommendations A-07-63 and -64 to address concerns that braking action reports provided by pilots and contaminant type and depth reports provided by airports were inherently subjective and thus not an optimal way to convey reliable information about runway conditions. Safety Recommendation A-07-63 asked the FAA to establish a minimum standard for “correlating an airplane’s braking ability to braking action reports and runway contaminant type and depth reports for runway surface conditions worse than bare and dry.” The FAA’s planned October 1, 2016, implementation of the RCAM would address the action requested in this recommendation because (1) airport operators will have standard criteria for reporting runway surface conditions along each one-third of a runway and (2) airframe manufacturers and others (supplemental type certificate holders and third-party suppliers) will provide corresponding performance data that air carriers will be able to use during flight planning and decision-making. As a result, Safety Recommendation A-07-63 remains classified “Open—Acceptable Response” pending the implementation of the RCAM and its use in reporting runway surface conditions. The implementation and use of the RCAM does not address the requested actions in Safety Recommendation A-07-64, which asked the FAA to do the following: Demonstrate the technical and operational feasibility of outfitting transport-category airplanes with equipment and procedures required to routinely calculate, record, and convey the airplane braking ability required and/or available to slow or stop the airplane during the landing roll. If feasible, require operators of transport-category airplanes to incorporate use of such equipment and related procedures into their operations. The FAA has sponsored the development and evaluation of these systems over the past several years and provided a briefing to the NTSB on September 6, 2016, about the status of this work.

[ANALYSIS] The first officer reported needing 7.5 hours of sleep per night to feel rested, so he had a cumulative sleep deficit of about 2 hours (see table 3) during that period. Research indicated that such sleep deficits could result in very small performance-related effects (Belenky and others 2003). However, these minor effects would likely have been offset by circadian factors that promote alertness during the late morning hours (when the accident occurred). In addition, at the time of the accident, the flight crew had not been on duty excessively long (about 6 hours including the accident flight), flown multiple trip legs (the accident flight was the second leg of the day), or been awake for a long time (about 7.5 hours for the captain and 7 hours for the first officer). 78 Although no deficiencies were noted in Delta’s weather documentation, the company’s meteorology department did not anticipate the low IFR conditions or 1/4-mile visibility during the period and the heavy snow that occurred immediately after the accident. 79 ATIS information Mike reported 1/2-mile visibility with snow and freezing fog. Title 14 CFR 121.613 prohibits the dispatch of a flight operating under IFR unless appropriate weather reports or forecasts, or any combination thereof, indicate that the weather conditions will be at or above the authorized minimums at the estimated time of arrival. The minimum visibility for an instrument approach to runway 13 at LGA was 1/2 mile; thus, at the time of dispatch, the forecasted visibility for LGA was at the required minimum visibility. NTSB Aircraft Accident Report 48 the runways were wet and had been sanded and deiced with solid chemical, which was not current given the report of thin, wet snow in the 0738 NOTAM. The flight was dispatched according to 14 CFR 121.195.80 Dispatch calculations indicated that the landing distance would be 5,995 feet (based on a planned landing weight of 132,500 pounds, flaps 40, a 15% safety margin for a wet runway, and maximum braking). The operational landing distance assessment, which included the use of maximum autobrakes with good braking action and reverse thrust (among other factors), was 6,200 feet—about 800 feet less than the available length of runway 13. Thus, the information that the flight crew had at the time of departure showed that the airplane would be able to safely stop on the runway. 2.1.3 En Route Preparations and Decision-making After the airplane departed ATL (about 0924), the flight crew continued to monitor the weather conditions at LGA and assessed the factors that could affect stopping performance. While in cruise flight at 0954:52, the flight crew consulted Delta’s contaminated runway crosswind limitations and recognized that the crosswind would be at Delta’s 10-knot limit if the runway braking action was reported to be medium/poor.

[ANALYSIS] The snow coordinator arrived on scene and told the controller at 1105:55 that the airplane was “leaking fuel on the left side of his aircraft heavily…his wing is ruptured.” Thirty seconds later (and about 3.5 minutes after the airplane came to a stop), ATC tower personnel activated the EANS, which reported “LaGuardia, Alert 3, all emergency vehicles respond. Alert 3, Delta 1086 MD-80, just east of runway 1-3, wing eruption, fuel is being leaked.”115 The airport’s ASDE-X showed ARFF vehicles leaving the fire station area about 1107 to respond to the accident. ARFF crews initially searched fenced areas at the approach end of runway 13 but did not locate the airplane there. Radio communications indicated that ARFF personnel were still unclear of the exact location and the severity of the accident as of 1110:12. ARFF personnel eventually saw the airplane on the embankment and arrived at the accident site about 1111:02, which was more than 8 minutes after the accident occurred. The initial information that airport operations personnel provided about the nature, location, and severity of the accident was somewhat ambiguous, which likely caused confusion for the local controller, who could not see the accident site. However, the controller did not actively seek additional information about the situation or effectively use the information available to him. In addition, airport operations personnel clearly requested at 1104:48 that the controller notify ARFF personnel about the accident, but the EANS was not activated for another 1 minute 37 seconds. Thus, the NTSB concludes that ARFF personnel would likely have arrived at the accident scene sooner if they had received more timely and precise information about the accident and its location. 2.5.2 Passenger Manifests The passenger count provided by the flight and cabin crews did not fully reflect the total number of passengers aboard the airplane. The first officer used the information in the final weight and balance report sent via ACARS to inform ATC tower personnel that 125 passengers were aboard the airplane. The lead flight attendant provided Port Authority personnel with the departure report, which also indicated that 125 passengers were aboard the airplane, and one of the flight attendants confirmed that passenger count to the LGA manager of airport operations. However, later in the day, LGA airport operations staff learned that 127 passengers were 115 A responding airport operations staff member who reported the accident to the controller used the term “3-4,” which had previously been used to describe the highest level of airplane emergency at LGA. However, ATC tower personnel used the correct term “alert 3” in the EANS notification. Thus, the initial incorrect terminology did not appear to cause any confusion for ATC tower personnel. NTSB Aircraft Accident Report 69 actually aboard the airplane. Two lap-held children were noted in the body of the departure report (with an adult’s name and seat location), but they were not included in the total passenger count. According to Delta, departure reports were generated from scanned passenger boarding passes, and lap-held children were not issued boarding passes.

[ANALYSIS] The FAA has consistently indicated that it would encourage, but not mandate, the use of airplane-based systems that determine runway surface conditions. However, given the success of CAST in effecting widespread voluntary introduction of safety improvements (including FAA-approved voluntary safety programs such as FOQA and the Aviation Safety Action Program), many operators might voluntarily equip their fleets with these systems if they are shown to be feasible, even without regulatory action by the FAA. Thus, if the FAA takes the 106 The NTSB understands that the FAA is still in the process of vetting and validating the calculations performed by this system and that work remains to determine how the calculated results are to be used by the company’s customers for operational decisions. NTSB Aircraft Accident Report 62 actions necessary to satisfy Safety Recommendations A-16-23 and -24, the intent of Safety Recommendation A-07-64 will also be satisfied without the FAA mandating the requested action. As a result, Safety Recommendation A-07-64 is reclassified “Closed—Acceptable Action/Superseded.” 2.4 Evacuation Issues 2.4.1 Evacuation Procedures After unexpectedly departing the paved runway surface, the airplane came to rest at a non-normal attitude with its nose over water and left wing damaged. The accident sequence also damaged the electrical systems, which resulted in the loss of the interphone and public address system as a means for communication. As a result, the two flight attendants in the aft cabin left their assigned locations and walked to the forward cabin (to obtain information from the flight crew and the lead flight attendant) while instructing passengers to stay seated and stay calm. The lead flight attendant left her assigned location to check on a passenger in the mid-cabin. However, the flight attendants were expected to stay near their assigned exit and remain in a state of readiness for an evacuation, which included assessing their exits to determine their status.107 Once the captain exited the flight deck, he met the flight attendants in the forward galley. The captain asked about the status of the forward and 2L door exits. The captain also asked whether the tailcone exit would also be available for an evacuation, and the flight attendant who had been seated at the tailcone replied “I don’t know.” Recognizing that no one was monitoring the aft cabin, one of the aft flight attendants walked back to her assigned position. A passenger stopped the flight attendant to tell her that a first responder was motioning to open the right overwing exits, to which she replied, “no, we need to wait until our captain instructs us to evacuate.” A first responder told the first officer, while he was in the cockpit, that everyone should evacuate the airplane via the right overwing exits due to fuel leaking from the left wing.

[ANALYSIS] Despite conducting annual inspections at LGA in February 2014 and February 2015 (before the accident) and in November 2015 (after the accident), the FAA’s ACSI was unaware that LGA was not using CFME during winter operations. He thought that the airport was using CFME for friction assessments because of the wording included in the Airport Certification Manual and letter of agreement with the ATC tower. Also, at the time of the accident, LGA’s computer-based training course for winter operations indicated that the airport was using friction measuring equipment during winter weather conditions. Further, although FAA ACSIs conduct occasional surveillance inspections in addition to annual inspections, the ACSI for LGA indicated that he had not been directed to observe active airport winter operations during his more than 5 years with the FAA. The NTSB is concerned that other ACSIs might not be fully aware of how an airport is assessing runway friction during winter weather operations. Therefore, the NTSB also recommends that, for airports certificated under Part 139, the FAA direct ACSIs to ensure, before or during the airports’ next scheduled annual inspection, that policies and procedures for friction measurement during winter operations are accurately and adequately described in the airports’ Airport Certification Manual and Snow and Ice Control Plan. 119 LGA also did not use CFME to perform a friction assessment of runway 13 after the accident, even after Delta’s request to do so. The NTSB notes that postaccident CFME runway friction assessments that are conducted immediately after an accident on a contaminated runway could provide valuable data to an investigation, especially if an airport operator had been using the equipment to analyze runway friction trends. NTSB Aircraft Accident Report 72 2.6.2 Runway Condition Reporting The last NOTAM before the accident, issued at 0903, indicated that 1/4 inch of wet snow was on runway 13/31. The last airport snow measurement before the accident (at 1051) showed that 0.4 inch of snow had accumulated during the last hour. On the basis of this snow accumulation and the time between the last runway clearing and the accident (27 minutes), the NTSB estimated that the runway was contaminated with about 1/4 inch of snow at the time that the accident airplane touched down. Although this snow amount was the same as that in the 0903 NOTAM, the NTSB is concerned that the NOTAM was 2 hours old at the time of the accident and had not been updated after the snow clearing operations. AC 150/5200-30C, which was current at the time of the accident, stated that NOTAMs describing the runway surface conditions must be “timely” because those conditions can change quickly as a result of winter weather conditions or the actions to mitigate those conditions. The AC also stated that runway condition reports needed to be updated any time a “change to the runway surface condition” occurred. Among other changes that would necessitate updated reports were weather events, chemical or sand applications, and plowing or sweeping operations.

[ANALYSIS] The AC also stated that runway condition reports needed to be updated any time a “change to the runway surface condition” occurred. Among other changes that would necessitate updated reports were weather events, chemical or sand applications, and plowing or sweeping operations. LGA’s chief operations supervisor stated that airport operations personnel did not routinely issue updated field condition reports after each snow clearing event. LGA’s aeronautical operations manager indicated that, as long as the snow removal teams could maintain the runway condition described in the most recent NOTAM, the NOTAM would remain in effect. The FAA ACSI for LGA also believed that a new NOTAM would not need to be issued under those circumstances. However, these postaccident statements seemed to contradict the FAA guidance’s in AC 150/5200-30C, which stated that airplane operations on runways should not be allowed after chemical or sand applications and plowing or sweeping operations until a new runway condition report was published to indicate the latest runway surface condition. The NTSB notes that the guidance in AC 150/5200-30C did not specifically describe what constituted a “timely” NOTAM and what types of “change to the runway surface condition” airport operations personnel needed to report. In addition, version D of the AC, which was issued in July 2016, does not provide clarification about these terms. If these terms were clarified, airport operations personnel could issue more effective NOTAMs, and flight crews could have more updated information regarding runway surface conditions. The NTSB concludes that the FAA’s airport winter operations safety guidance is not sufficiently clear about the timing and need for updated runway condition reports, which could result in flight crew uncertainty about possible runway contamination. Therefore, the NTSB recommends that the FAA revise AC 150/5200-30D to provide more precise guidance regarding (1) the need to issue NOTAMs in a timely manner and (2) the specific changes to runway surface conditions that would prompt the issuance of updated NOTAMs. During a December 2015 meeting between LGA airport operations staff and the chief pilots for air carriers that operate at the airport, LGA staff indicated that it would update NOTAMs during winter operations each hour, regardless of whether runway conditions changed, and as necessary between hourly reports. The NTSB believes that this approach can help ensure that pilots receive timely and accurate information about the runway conditions at LGA.

[ANALYSIS] Flight and cabin crewmembers involved in an accident or incident should be able to provide emergency responders with an accurate passenger count (including lap-held children) upon exiting the airplane and without contacting company personnel for further information. Therefore, the NTSB recommends that the FAA clarify guidance to all Part 121 air carriers to reinforce the importance of (1) having precise information 116 More information about this incident is available in the Survival Factors Specialist’s Factual Report in the incident docket. 117 As a result of the ValuJet flight 592 accident, the NTSB issued Safety Recommendation A-97-77, which asked the FAA to “instruct principal operations inspectors to review their air carriers’ procedures for manifesting passengers, including lap children, and ensure that those procedures result in a retrievable record of each passenger’s name.” The NTSB classified this recommendation “Closed—Acceptable Action” on July 23, 1999. NTSB Aircraft Accident Report 70 about the number of passengers aboard an airplane, including lap-held children, and (2) making this information immediately available to emergency responders after an accident to facilitate timely search and rescue operations. 2.6 Airport Issues 2.6.1 Runway Friction Measurement Policies Many airports use CFME or decelerometers during winter conditions to assess the effectiveness of runway clearing operations.118 LGA’s Airport Certification Manual stated that the airport used CFME to conduct friction readings when conditions required trend analysis on a frozen or contaminated surface. Also, Port Authority’s letter of agreement with the LGA ATC tower stated that airport operations staff could conduct friction assessments when conditions might result in degraded runway surface friction and that such assessments could be conducted using CFME. However, LGA’s chief operations supervisor stated that the airport’s two CFME vehicles were not used during snow removal operations and that the vehicles were only used to assess runway friction related to rubber removal. LGA had not been using CFME during winter operations since a change in Port Authority policy that was effective in November 2011. The policy was based on information in a letter from the FAA’s Office of Airport Safety and Standards, which responded to Port Authority’s questions about the airport winter operations guidance in AC 150/5200-30C. The letter confirmed that the FAA did not require airports to conduct friction surveys in winter conditions but recognized that “operational [friction] testing under winter conditions can be a valuable tool to airport operators in providing information on changing runway conditions.” This information was consistent with the FAA’s guidance in AC 150/5200-30C, which stated that “some airport users still consider runway friction measurement values to be useful information for tracking the trend of changing runway conditions.” The letter also confirmed that the FAA no longer recommended providing friction measurements (Mu values) to pilots but advised that providing friction measurement reports to other interested parties for the purpose of trend analysis was permitted.

[CONCLUSIONS > FINDINGS] However, the records were inadequate to use for trend analysis or evaluation of an individual's performance during training. 5. Although air traffic control was not a factor in this accident, because of procedural shortcomings, windshear advisory information was not provided to the flightcrew in a timely fashion. 6. The practice of displaying only the centerfield wind on the low level windshear alert system limited the amount of information the controller had available to him to issue to the flightcrew. 7. The LL WAS system operated within acceptable limits at the time of the accident. 8. NWS and American Airlines weather information provided to the flightcrew of AAL102 was timely and substantially accurate. 9. At touchdown, flight I 02 was subjected to cross-track winds of about 15 knots that may have been increasing, with gusts about 5 knots above the steady winds. 106 10. The wind gusts of 25 to 32 knots recorded on the centerfield anemometer were not a factor in the accident; however, wind speed surges of similar strength could not be ruled out entirely based on meteorological data. 11. No microbursts or hazardous low level windshears affected the airplane at the time of the landing. 12. A line of moderate to heavy showers and thunderstorms was crossing runway 17L as AALI 02 touched down. 13. The captain failed to compensate for moderate crosswinds from the right, allowing the airplane to weathervane and drift off the right side of the runway with minimal rudder commands, inappropriate tiller nosewheel steering commands, and lack of forward pressure on the control column. 14. The evacuation of the passengers, although made difficult by the fire and the nose-down left roll final resting attitude of the airplane, in mud, was handled in an expeditious and professional manner. 15. The emergency lighting did not operate properly because the emergency overhead lighting system battery packs were found to be out of sequence. This condition resulted in enough electrical power to indicate that the system was fully charged on the flight engineer's console but insufficient power to operate the overhead emergency lighting system for a specified 5 minutes. The manufacturer's instructions did not describe the importance of properly sequencing the batteries in each pack. 16. Runway 17L-35R was worn to a "maintenance planning" level. However, the majority of the runway's coefficient of friction was found to be within prescribed advisory circular guidelines. 17. There is inadequate FAA oversight of the runway friction measurement at U.S. airports. 107 18. The emergency response of the DFW ARFF was exceptionally good. 19.

[CONCLUSIONS > FINDINGS] Runway 17L-35R was worn to a "maintenance planning" level. However, the majority of the runway's coefficient of friction was found to be within prescribed advisory circular guidelines. 17. There is inadequate FAA oversight of the runway friction measurement at U.S. airports. 107 18. The emergency response of the DFW ARFF was exceptionally good. 19. There was no evidence of hydroplaning on the runway or reverted rubber on the airplane's tires. 20. The broken nose landing gear steering cables were a result of the accident. They were not broken before the collapse of the nose landing gear. 21. Two of 32 reverser cascades on the center or No. 2 engine were found not to be in the proper configuration. The calculated misdirected force could be counteracted by about I degree of rudder deflection. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of the accident was the failure of the captain to use proper directional control techniques to maintain the airplane on the runway. 108 4. RECOMMENDATIONS As a result of the investigation of this accident, the National Transportation Safety Board makes the following recommendations: --to the Federal A vi at ion Administration: Review the pilot training recordkeeping systems of airlines operated under FAR Parts 121 and 135 to determine the quality of information contained therein, and require the airlines to maintain appropriate information on the quality of pilot performance in training and checking programs. (Class II, Priority Action) (A-94-24) Amend the ATC handbook, 7110.65, Chapter 3, "Airport Traffic Control - Terminal," Section 1, General: paragraph 3-8, "Low Level Windshear Advisories," to state that tower controllers should issue the LLWAS advisory, "Low Level Windshear Advisories in Effect," whether or not the facility is equipped with an ATIS.

[CONCLUSIONS] CONCLUSIONS FInding ........ es ceesseccesscceessceeceseceeescccesseeeeessseesesssecessosssesesssseesessseesenee 105 Probable Cause... ssssssccsesscseccsssnssscsssnsssessesesssscceecsesssssseesessessseees 107 RECOMMENDATIONS ......... cc ccsssccessscssrecsssccessececssnecessacessneessnees 108 APPENDIXES Appendix A--Investigation and Hearing ............cccsscccccssssssseseeseeseeees 111 Appendix B--Personnel] Information...............cccsssssssssccsecteeesseseseseesees 112 Appendix C--Cockpit Voice Recorder Transcript ...........csscssssssseeeees 116 Appendix D--DFW Tower/TRACON Transcript ..........:ccssceesseeeseeees 149 Appendix E--Antiskid System........cccescsccsssccsssscessesceserseseceneesonenees 163 Appendix F--Selected Weather Radar Data .............::ccccsssecsesseeeeeensees 165 vi EXECUTIVE SUMMARY On April 14, 1993, about 0659:43 central daylight time, American Airlines flight 102, a McDonnell Douglas DC-10-30, departed runway 17 left, following landing at Dallas/Fort Worth International Airport, Texas, after a nonstop, overnight flight from Honolulu International Airport, Hawaii. It was raining at the time of the landing, and there were numerous thunderstorms in the area. There were 189 passengers, 3 flightcrew members and 10 cabincrew members aboard the airplane. Two passengers received serious injuries, and 35 passengers, 1 flightcrew member, and 2 cabincrew members received minor injuries during the evacuation of the airplane. The airplane sustained substantial damage. The National Transportation Safety Board determines that the probable cause of the accident was the failure of the captain to use proper directional control techniques to maintain the airplane on the runway. The safety issues in this report focused on weather conditions affecting the flight, flightcrew and air traffic control training and procedures, airplane emergency evacuation lighting, and runway maintenance. Safety recommendations concerning these issues were addressed to the Federal Aviation Administration, Dallas/Fort Worth International Airport, and American Airlines, Inc.

[ANALYSIS] The systemactivating cockpit switch was found by investigators in the "standby" position, and the switch at the 1-L flight attendant station was found in the "off' position. No deficiencies were found in either the subsystem or in the components on the airplane. The cockpit and 1-L flight attendant activation switches were found functional. In addition, voltage and impedance tests conducted on the wiring of each subsystem indicated no deficiencies. Additional testing was done under Safety Board supervision on both subsystems at the facilities of their manufacturers. This inspection and testing determined that the Gulton systems battery pack had been reassembled improperly during maintenance. There was no evidence of improper assembly by the manufacturer. The individual battery packs (constituting 24 batteries for each of the four battery charging units) are required by maintenance scheduling to be replaced in the same sequence as they were previously installed on the airplane. Three of the four battery packs were found to have been configured improperly. With the improper configuration, enough electrical power would have been provided to indicate an "up system" in battery tests but not enough power to energize the emergency lights in the actual emergency. Upon receiving this information, American Airlines issued an Engineering Service Order and initiated an inspection of battery stores in all airplanes that might have been affected by the configuration. 2.8 Landing Runway Surface Conditions It is apparent that the surface texture of 17L-35R had deteriorated with use, or as a result of high levels of jet traffic and weather-related erosion. FAA guidance, as stated in FAA Advisory Circular (AC) 150/5320-12B, addresses runway wear. Although, by definition, "maintenance planning" for this runway was called for, the friction levels of the majority of the runway fell within acceptable levels for airplane operations. Furthermore, as described below, there is no evidence that the airplane entered hydroplaning on the runway or that traction was significantly reduced because of the condition of the runway. 101 The investigation foun·d a buildup of rubber at the approach end of 17L that showed a coefficient of friction below the FAA minimwn standard. According to airport records, for the past 3 years, rubber removal was conducted at 4- and 8-month intervals. There was an average of 261 landings on 17L each day. FAA guidance suggests a rubber removal frequency of every 2 months for runways with a frequency of turbojet landings of more than 210 per day. The Safety Board concludes that DFW should monitor the runways more frequently, and, if necessary, remove the rubber buildup on all runways, in accordance with the referenced directive. However, because the accident flight landed long, the airplane did not traverse the areas where rubber buildup was found. Although this buildup needs to be corrected, it did not contribute to the loss of directional control on the runway. The FAA provides guidance in advisory circulars for runway friction measurement and runway maintenance._ However, there is no formal requirement for FAA oversight of airports regularly performing friction measurements.

[ANALYSIS] However, because the accident flight landed long, the airplane did not traverse the areas where rubber buildup was found. Although this buildup needs to be corrected, it did not contribute to the loss of directional control on the runway. The FAA provides guidance in advisory circulars for runway friction measurement and runway maintenance._ However, there is no formal requirement for FAA oversight of airports regularly performing friction measurements. In addition, there are no formal requirements for the FAA to regularly inspect certificated airports to ensure that they have adequate friction measurement or, if necessary, rubber removal programs. The Safety Board has addressed the subject of runway friction since 1973 when Safety Recommendation A-74-119 was issued to the FAA to: Amend appropriate regulations and procedures to establish an alerting service to advise pilots of hydroplaning probabilities before and during the landing approach. Such an advisory system would entail (1) A runway slipperiness rating and runway contamination monitoring program; and (2) The use of measuring devices and associated charts to correlate rainfall, wind direction, and velocity, with runway gradient and water depth on the runway surface. Safety Recommendation A-74-119 was superseded by the recommendations issued with the accident report on the January 23, 1982, World Airways accident at the Boston-Logan Airport. The applicable recommendations in the Boston-Logan World Airways report were superseded by the recommendations in the report on the October 25, 1986, accident at the Charlotte-Douglas airport-- specifically A-87-110. In all, 19 safety recommendations have been issued regarding runway friction and friction measurement. 102 As a result of the Safety Board investigation of the Piedmont Airlines Boeing 737 accident at Charlotte-Douglas International Airport, North Carolina, on October 25, 1986, the Safety Board issued Safety Recommendation A-87-110 which recommended that the FAA: During annual inspections of full certificate airports, emphasize the identification of deficient runway conditions and use approved friction measuring devices to measure the dry runway coefficients of friction. Encourage the airport operator to correct ( or provide appropriate notice to users) runway conditions that do not meet criteria recommended in advisory circular 150/5320-12A. Following previous correspondence, on December 12, 1992, the FAA replied that it agreed with the safety recommendation and has revised AC 150/532012B to include guidance and procedures for the design and construction of skidresistant pavement, pavement evaluation with or without friction equipment, and maintenance and high skid-resistant pavements. As a result of that letter, on March 26, 1993, the Safety Board classified Safety Recommendation A-87-110 as "Closed--Acceptable Action." However, as a result of the investigation of the accident involving AALI 02, the Safety Board believes that the FAA should take a more assertive role in overseeing airport runway friction measurement programs.

[CONCLUSIONS] Existing Federal Aviation Administration pilot flight and duty time regulations 21. permitted the long and demanding day experienced by the accident pilots, which resulted in their fatigued condition and degraded pilot decision‑making. The additional responsibilities and task demands involved in providing operating 22. experience and performing related check airman functions likely aggravated the effects of fatigue for the captain/check airman. The pilots could have made a more informed landing decision if they had monitored 23. the current (updated every minute) and unambiguous Cherry Capital Airport (TVC) weather information that was continuously available to them through the TVC automated surface observing system broadcast. The Cherry Capital Airport operations supervisor’s use of ambiguous and unspecific 24. radio phraseology when providing braking action information likely affected the captain’s decision to continue the approach; an unambiguous runway surface condition report would have provided the pilots with more accurate and useful information to factor into their landing decision. Incorporation of minimum safe operating limits for runway surface conditions into 25. an airport’s snow and ice control plan would ensure that airport operations personnel prohibit air carrier operations on any runway if, in their estimation, the braking action on that runway is unsafe. Probable Cause 3.2  The National Transportation Safety Board determines that the probable cause of this accident was the pilots’ decision to land at Cherry Capital Airport (TVC), Traverse City, Michigan, without performing a landing distance assessment, which was required by company policy because of runway contamination initially reported by TVC ground operations personnel and continuing reports of deteriorating weather and runway conditions during the approach. This poor decision-making likely reflected the effects of fatigue produced by a long, demanding duty day and, for the captain, the duties associated with check airman functions. Contributing to the accident were 1) the Federal Aviation Administration pilot flight and duty time regulations that permitted the pilots’ long, demanding duty day and 2) the TVC operations supervisor’s use of ambiguous and unspecific radio phraseology in providing runway braking information. National Transportation Safety Board A I R C R A F T Accident Report 56 Recommendations 4. New Recommendations 4.1  As a result of this investigation, the National Transportation Safety Board makes the following recommendations to the Federal Aviation Administration: Emphasize with principal operations inspectors the importance of conducting timely postaccident drug and alcohol testing. (A-08-40) As part of the Takeoff/Landing Performance Assessment Aviation Rulemaking Committee, address the need for initial training on the rationale for and criticality of conducting landing distance assessments before landing on contaminated runways. (A-08-41) Issue a CertAlert to all 14 Code of Federal Regulations Part 139 certificated airports that describes the circumstances of this accident, emphasizes the importance of specific and decisive radio communications, and urges airports to ensure that those criteria are being met in all airfield radio communications. (A‑08-42) Require all 14 Code of Federal Regulations Part 139 certificated airport operators to include in their airport’s snow and ice control plan absolute criteria for type and depth of contamination and runway friction assessments that, when met, would trigger immediate closure of the affected runway to air carrier operations.

[ANALYSIS] Further, postaccident interviews and CVR evidence indicated that the pilots were aware of these required procedures. Among other OE procedural and operations issues discussed during the flight from MSP to TVC, the captain specifically reviewed uncontrolled airport operations and procedures with the first officer. However, CVR data showed that the crew monitored the TVC ASOS weather information on only one occasion, about 30 minutes before they landed. They did not obtain a more current observation as they neared the airport and heard transmissions from the airport operations supervisor indicating that the conditions were deteriorating. Therefore, the Safety Board concludes that the pilots could have made a more informed landing decision if they had monitored the current (updated every minute) and unambiguous TVC weather information that was continuously available to them through the TVC ASOS broadcast. Airport and Runway Condition Reports and Ground 2.4.2  Personnel Phraseology In accordance with published procedures, after tower closing time the pilots communicated directly with TVC airport operations personnel on the CTAF regarding the timing of their arrival, snow removal activities, and the airport/runway conditions. The airport operations personnel are the source of runway and field condition reports issued by the control tower during operating hours, which are distributed via NOTAMs; therefore, the accident flight crew had direct access to equivalent runway and field condition information. The Safety Board notes that early in the airplane’s descent, the TVC airport operations supervisor provided the following precise runway condition report to the accident captain, “I’ve [.40+ on] runway two eight. I’ve got thin, wet snow [over] patchy thin ice.” However, subsequent phraseology used by the TVC airport operations supervisor during radio transmissions to the accident pilots was conversational and imprecise (for example, the phrases “comin’ down good,” “fillin’ in real hard,” and “probably nil”) and therefore subject to possible misinterpretation. During the last 20 minutes of the accident flight, the TVC airport operations supervisor made several radio transmissions to the accident pilots regarding snow removal operations, subsequent snowfall, and accumulation of snow on the landing runway. Additionally, about 0040:53, the pilots clearly heard the second of two transmissions issued by the TVC airport operations supervisor that described the braking action on runway 28 as “probably nil.”74 However, when the captain immediately asked, “are you saying it’s nil?” to confirm the nil braking report, the TVC airport operations supervisor vacillated and eventually downplayed his nil report, saying he had not performed a field report and did not know “what it’s doin’ now.” When queried by the captain, the TVC 74  As previously stated, the pilots likely did not hear the first nil braking report because of a simultaneous radio transmission. Analysis National Transportation Safety Board A I R C R A F T Accident Report 51 airport operations supervisor estimated the runway snow depth to be “close to” 1/2 inch.

[ANALYSIS] The pilots received updated weather information from the company via ACARS about 45 minutes before landing. This updated information indicated that the winds at TVC were still favorable for landing, and the captain advised the passengers that “it looks like we’re gonna have no problems gettin’ in [to TVC] this evening.” Pilot Actions and Decision-Making During the Approach 2.2.1  Landing Distance Assessments 2.2.1.1  CVR and postaccident interview evidence indicated that the pilots’ concerns during the flight appeared to be primarily related to the TVC wind conditions, perhaps because that was the critical factor in the airplane’s dispatch. Although the CVR recorded the captain mentioning the possibility of diverting to DTW late in the approach (about 1 minute before touchdown), the pilots exhibited limited concern regarding the runway surface condition. About 37 minutes before they landed, the pilots listened to the TVC ASOS broadcast for updated weather information and runway surface condition information. This ASOS information indicated winds out of 040º at 7 knots and visibilities of 1 1/2 miles in light snow. This was the only TVC ASOS broadcast the pilots listened to before their arrival at TVC. However, TVC ground operations personnel provided the pilots with updated weather and runway surface condition information on several occasions as the airplane neared the airport. Evidence indicates that the runway surface conditions at TVC deteriorated further because of increasing snowfall during the last 15 minutes of the accident flight. The TVC airport operations supervisor provided runway surface condition information to the accident pilots both before (“forty plus MU”65 with “thin wet snow over patchy thin ice”) and during the vectoring stages of the approach (for example, “it’s comin’ down pretty good,” “this is fillin’ in pretty quick,” and “it’s fillin’ in real hard”). Consistent with this information, the CVR recorded the captain commenting that he expected to land on a contaminated runway. For example, the captain stated: “there’s snow removal on the field yet they’re showing forty or better sounds like a contaminated…runway to me” at 0029:10.5; “with contaminant, more than likely” at 0033:50.9; and “snowing hard” at 0034:09.3. (ASOS reports, which the pilots had not obtained, also showed that light snow increased to moderate snow about 0030; then, about 0040, increased to heavy snow with visibility of 1/4 mile.) In December 2006, Pinnacle incorporated procedures into the company’s OpSpecs requiring its pilots to perform landing distance assessment procedures consistent with guidance contained in the FAA’s SAFO 06012.

[FINDINGS] 53 Conclusions 3. Findings 3.1  The pilots were properly certificated and qualified under Federal regulations and 1. Pinnacle Airlines training requirements. No evidence indicated any medical conditions that might have adversely affected the pilots’ performance during the accident flight. The accident airplane was properly certificated and was equipped and maintained 2. in accordance with industry practices and was within weight and center of gravity limits. The investigation revealed no evidence of any failure or anomaly of the airplane’s 3. powerplants, structures, or systems (including the airplane’s deceleration devices, such as brakes, antiskid devices, and thrust reversers) that would have affected the airplane’s performance during the accident landing. Based on the system designs and runway conditions, it is likely that the airplane’s 4. braking and antiskid systems were performing to their maximum effectiveness. Although, at its original departure, time Pinnacle dispatchers could not dispatch the 5. accident flight because of strong winds in the Cherry Capital Airport (TVC) forecast, an amended forecast issued by Northwest Airline’s meteorology department (and reflected in a subsequent National Weather Service-issued forecast) predicted more favorable wind conditions (as well as higher ceilings and improved visibility in light snow) at TVC and thus met the required criteria for the flight’s dispatch. The services provided by the air traffic control (ATC) system did not affect the outcome 6. of the flight. Information commonly provided by ATC (for example, weather and runway surface condition reports) was available to the flight crew, and its availability was unaffected by the air traffic control tower’s closure before the flight’s arrival at Cherry Capital Airport. Cherry Capital Airport’s snow removal operations and runway surface condition 7. assessment activities were conducted in accordance with the airport’s Federal Aviation Administration-approved snow and ice control plan. Considering the severe winter weather and the relatively intact condition of the 8. airplane, the captain’s decision to deplane the passengers using Pinnacle’s “expeditious deplaning” procedures was appropriate. Although there were no reported injuries resulting from this accident, had a 9. postaccident fire occurred, the delay in aircraft rescue and firefighting response could have adversely affected the safety of passengers after the accident. Conclusions National Transportation Safety Board A I R C R A F T Accident Report 54 The forward-looking infrared equipment installed in the aircraft rescue and firefighting 10. (ARFF) vehicle did not help the firefighter locate the accident airplane; however, improved crash detection and location equipment would likely have facilitated a more timely ARFF response. Although there is no reason to believe the pilots’ performance was affected by alcohol, 11. the failure of the airline to perform required postaccident alcohol tests prevents a definitive statement on the issue. Even though there was initially some uncertainty as to whether the Cherry Capital 12. Airport runway overrun was an accident or an incident, it would have been prudent for Pinnacle to comply with the drug and alcohol testing regulations as if the overrun were to be classified as an accident. The pilots failed to perform the landing distance assessment that was required by 13.

[ANALYSIS] Analysis National Transportation Safety Board A I R C R A F T Accident Report 51 airport operations supervisor estimated the runway snow depth to be “close to” 1/2 inch. CVR-recorded communications between the pilots indicated that the captain was initially concerned about the TVC airport operations supervisor’s “probably nil” braking report but felt more confident about landing after hearing the contaminant depth estimate of 1/2 inch. The FAA recommends that airports use the AIM chapter titled, “Radio Communications, Phraseology, and Techniques” as a source for related airport training materials and procedures. This AIM chapter emphasizes the importance of precision, conciseness, and proper radio technique in successfully communicating by radio and includes examples of proper phraseology and radio techniques. A review of TVC training materials for operations personnel regarding communications and airport familiarity indicated that the materials were consistent with and referenced the AIM information. Further, TVC records indicate that all ground personnel on duty the night of the accident, including the airport operations supervisor, had successfully completed the required training. Because the airport operations supervisor had completed the required training and was also familiar with appropriate phraseology from his pilot training, it is not clear why he failed to provide specific and decisive information at all times on the night of the accident. However, it is likely that TVC airport operations supervisor’s reticence to confirm his “probably nil” braking report when the captain tried to confirm it was a factor in the pilots’ decision to continue the approach. The Safety Board concludes that the TVC airport operations supervisor’s use of ambiguous and unspecific radio phraseology when providing braking action information likely affected the captain’s decision to continue the approach; an unambiguous runway surface condition report would have provided the pilots with more accurate and useful information to factor into their landing decision. Therefore, the Safety Board believes that the FAA should issue a CertAlert to all 14 CFR Part 139 certificated airports that describes the circumstances of this accident, emphasizes the importance of specific and decisive radio communications, and urges airports to ensure that those criteria are being met in all airfield radio communications. In its report on the Southwest Airlines flight 1248 accident, the Safety Board issued Safety Recommendation A-07-62, asking the FAA to do the following: Develop and issue formal guidance regarding standards and guidelines for the development, delivery, and interpretation of runway surface condition reports. The FAA responded to Safety Recommendation A-07-62 on January 8, 2008, stating that the ARC (development of which was announced by the FAA on December 6, 2007) would also provide the FAA with advice and recommendations regarding “establishment of standards for runway surface condition reporting and minimum surface conditions for continued operations.” On June 12, 2008, the Safety Board classified Safety Recommendation A-07-62 “Open—Acceptable Response.”

[ANALYSIS] The services provided by MSP ARTCC were in accordance with all FAA directives. Separation, instrument approach vectoring and clearance, and additional services were all complete and correct. The flight arrived after the TVC ATCT had closed for the night; however, the pilots required no ATC traffic information or separation services because there was no other traffic in the area at the time of arrival (records indicate that the most recent traffic in the area was about 3 hours earlier). The Safety Board concludes that the services provided by the ATC system did not affect the outcome of the flight. Information commonly provided by ATC (for example, weather and runway surface condition reports) was available to the flight crew, and its availability was unaffected by the ATCT’s closure before the flight’s arrival at TVC. TVC airport operations personnel performed and completed snow removal and deicing operations based on the accident airplane’s revised estimated arrival time of 0037. Friction measurements completed on runway 28 after those operations (about 15 minutes Analysis National Transportation Safety Board A I R C R A F T Accident Report 38 before the airplane touched down, as prescribed in the airport snow removal plan) were better than .40 MU. However, moderate-to-heavy snow fell after the snow removal operations were completed, and the runway conditions deteriorated. Ground personnel estimated that the snow depth had increased to nearly 1/2 inch before the airplane landed. Based on the airport’s snow and ice control plan, this depth of accumulation nearly, but did not quite, reach the 1/2-inch level needed to trigger additional snow removal activities. Therefore, the Safety Board concludes that TVC’s snow removal operations and runway surface condition assessment activities were conducted in accordance with the airport’s FAA-approved snow and ice control plan. Postaccident interviews and documentation indicated that after the airplane came to a stop, the captain evaluated its condition and considered various methods of deplaning the passengers. The captain recognized that if he commanded an immediate emergency evacuation, the passengers would be safely away from the airplane; however, they would have been exposed to severe winter weather conditions at night. Because the pilots’ postaccident evaluation of the airplane (which included an external inspection) revealed no indication of fire risk, the captain elected to follow Pinnacle’s “expeditious deplaning” procedures. He kept the passengers on board the airplane until emergency response/transport personnel arrived at the airplane. Postaccident reports indicate that the resultant egress through the left front cabin door and stairs was orderly; there were no injuries reported. The Safety Board concludes that, considering the severe winter weather and the relatively intact condition of the airplane, the captain’s decision to deplane the passengers using Pinnacle’s “expeditious deplaning” procedures was appropriate.

[ANALYSIS] Analysis National Transportation Safety Board A I R C R A F T Accident Report 52 Runway Closure Procedures in Snow and Ice Conditions 2.4.3  In 2005, the FAA Great Lakes Region sent correspondence to all Part 139 airport operators in the region regarding operations during snow and ice conditions. This letter stated that airports must implement procedures for closing any pavement available to air carriers when braking action/friction values reach an unsafe value (the equivalent of nil braking action). TVC airport personnel discussed this issue at a snow plan meeting held September 22, 2006. However, at the time of the accident, TVC’s snow and ice control plan did not specify criteria that would result in airport personnel closing a runway and/or the airport. If TVC’s snow and ice control plan had incorporated such criteria, it is likely that the TVC airport operations supervisor would have, given his determination that the braking action was nil, closed the runway to air carrier operations before the accident flight arrived, forcing the pilots to take alternate action. (After the accident, TVC received operational criteria information from the air carriers, and the airport now restricts air carrier operations when MU values of .27 or less are measured or when nil braking action is reported by pilots or TVC ground operations personnel.) Therefore, the Safety Board concludes that incorporation of minimum safe operating limits for runway surface conditions into an airport’s snow and ice control plan would ensure that airport operations personnel prohibit air carrier operations on any runway if, in their estimation, the braking action on that runway is unsafe. Therefore, the Safety Board believes that the FAA should require all 14 CFR Part 139 certificated airport operators to include in their airport’s snow and ice control plan absolute criteria for type and depth of contamination and runway friction assessments that, when met, would trigger immediate closure of the affected runway to air carrier operations. Friction assessments should be based on pilot braking action reports, values obtained from ground friction measuring equipment, or estimates provided by airport ground personnel. National Transportation Safety Board A I R C R A F T Accident Report 53 Conclusions 3. Findings 3.1  The pilots were properly certificated and qualified under Federal regulations and 1. Pinnacle Airlines training requirements. No evidence indicated any medical conditions that might have adversely affected the pilots’ performance during the accident flight. The accident airplane was properly certificated and was equipped and maintained 2. in accordance with industry practices and was within weight and center of gravity limits.

[ANALYSIS] Even though there was some uncertainty as to whether the runway overrun was an accident or an incident, it would have been prudent for Pinnacle to comply with the drug and alcohol testing regulations as if the overrun were to be classified as an accident. After another recent accident,64 pilots tested negative for alcohol and drugs. However, the alcohol testing was delayed until about 3 hours after the accident and no records stating the reasons for the delay were prepared by the air carrier—Shuttle America—(as required by 14 CFR Part 121 Appendix J) nor were such records requested by an FAA representative. Timely testing for alcohol after an accident is necessary to evaluate any safety factors related to alcohol impairment or to eliminate them from further consideration. Although there was no evidence that alcohol was a factor in either recent accident, it cannot be conclusively ruled out; further, there is evidence that administration of required testing was not conducted or enforced strictly. Therefore, the Safety Board believes that the FAA should emphasize with POIs the importance of conducting timely postaccident drug and alcohol testing. The safety issues discussed in this analysis include the pilots’ actions and decisionmaking during the approach, landing, and landing roll; pilot fatigue and line check airman duty time regulations; weather and field condition information and ground operations personnel communications; and criteria for runway closures in snow and ice conditions. Pilot Performance, Actions, and Decision-Making 2.2  During the Flight During its investigation, the Safety Board evaluated the pilots’ actions and decisions during the flight, including their decision to land at TVC, their awareness of/attention to the weather conditions at TVC, and their actions during the landing roll. The Safety Board’s review of CVR evidence indicated that, consistent with the captain’s performing 64  National Transportation Safety Board, Runway Overrun During Landing, Shuttle America, Inc., doing business as Delta Connection Flight 6448, Embraer ERJ-170, N862RW, Cleveland, Ohio, February 18, 2007, Aircraft Accident Report NTSB/AAR-08/01 (Washington, DC: NTSB, 2008). Analysis National Transportation Safety Board A I R C R A F T Accident Report 40 OE duties, the pilots’ conversation during the flight largely focused on operational and procedural issues, including the inclement weather (snow and strong winds). Postaccident interviews and CVR evidence showed that the pilots had been operating in inclement weather conditions with snow, wind, and turbulence all day and expected to encounter similar weather conditions at TVC. The pilots received updated weather information from the company via ACARS about 45 minutes before landing.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] This was a potentially dangerous situation. This accident record revealed several examples of assumptions made by involved personnel that “someone else" verified the accuracy and/or the currency of information being used. Assumptions wete also made that the “other person was using the same data when communicating about this operation. In this case, as in most accidents, there was a chain of events which culminated with the accident. If that chain had been broken at any point the accident could have been prevented The events in this chain are usually designated as contributing factors and normally they can be eliminated, thus breaking the chain, by adhering without exception or deviation to established rules and procedures, The Board believes that any phase of an operation, e.g, providing the facilities, preparing of charts, planning and preparing the papér work, or the briefing of crewmembers, is just as importante and vital, and must receive the same meticulous attention to deiail, as the operation of the aircraf: itself. The captain vecepted the planning for a departure from 28L and he assumed the fligh: controller had checked the availability of the runway before he prepared the paperwork for the fight. The captain did not check: the airport conditions although he could have done so. His past experiences relating to the validity of the information provided by a flight contioller probsbly negated any need to check this data. The Board believes it would be appropriate for fiight crews to verify the airport conditions before they leave the dispatch office in the same manner that they verify weight and balance flight planning, etc. _ After the crew had boarded the aircraft and had gone through the routine of the pre-start checks, wherein the “Bugs” (V reference speeds) predicated on a 10° flaps setting for takeoff were set on the captain’s and first officer's respective airspeed indicators, the aircraft was pushed back from the gate, the engines were started and taxi clearance was requested. it was during the above sequence that the first officer, while listening to the ATIS broadcast, became aware of the closing of Runway 28L, the prevailing northwest wind, and that Runway 91 was being used for departures. A combination of the wind information and the aircraft weight caused the first officer to request Runway 28R when the tower cleared the flight to taxi to Runway 01K. The captain heard the clearance and the request for 28R and. at that time, told his first officer he wanted 28L. It was obvious from this interchange that, at that time. the captain was unaware of the closure of 28L. The radio request to PANOP, “... like to use 28R - would you check that along with us?”, was a prudent step.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] However, at the time of the accident, there had been no tests performed to detcrmine whether or not this point was the minimal required distance from the freeway for the jet blast to be diminished or deflected. The general attitude of the airpore personnel seemed to be that since no one had complained of ihe shortening of the runway there was no need to examine the adequacy or inadequacy of the change. The airport operations personnel had, in accordance with their procedurcs, published and disseminated to the appropriate distribution points an Airport NOTAM about an operational restriction to Runway O1R. The distribution points, such as the FSS representative of the FAA, were then expected by the airport personnel, to disseminate the message in accordance with the FAA procedures. The airport personnel’s procedures for initiating action and disseminating information were adequate, How. ever, they had no method to insure that the information they initiated was actua'ly transmitted, or, if it was transmitted, in what form. They did have a second system of communicating information to local users, the telephonic updating of ficld conditions, but, as has been demonstrated, Pan American flight controllers did not use this system. There was no way of ascertaining who received the information placed on the telephonic briefing tape and no way to determine when any information could safely be deleted from the tape. The airport authority is commended for having such a tool available for use. However, the briefing tape cannot be relied t + aR A, li Rae NE LRN anal ar ed Reser Ik NO RNS MLD an ear ote 1 t }} ; 5 i : upon or accepted as a panacea to the communication of important or necessary informa. tion. The FAA did not disseminate the airport information delivered to them in accordance with their own rules, Several conditions (c.g. Runway 01 restriction, Runway 28L closure and OLR partial closure), all of which qualified as information to be disseminated under the NOTAM criteria, were omitted from the system. Also, information submitted as NOTAM information by appropriate organizetions was reclassified by FSS personnel, sometimes crroncously, without informing the criginator. NOTAM’s are transmitted on the “Service A” and AIRAD’s on the “Service B” teletype circuits. However, all users do not have both of these circuits. Therefore, information improperly classified and put ont as AIRAD information would not reach users who do not have “Service B.” Since the information about the operating restriction on Runway OR for the 3-747 was reclassificd as an AIRAD, and since Pan American did not have “Service B” at San Francisco, this may explain why the Pan American flight control personnel were unaware of the 1,100-font restriction for B-747’s.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] The actual length of Runway 28R was not known to those required to know and to publish it. Two different lengths were given by representatives of the airport authority and a third length was listed in the airport diagram attached to the FAA Airport Master Record (FAA FORM 5010-1) dated April 1, 1971. The lettcr from the Chicf, Photogrammetry Division, National Oceanic and Atraospheric Administration, dated November 6, 1970, and another airport diagram attached to the FAA Form 5010-1, dated April 30, 1970, contained additional differing lengths for the same runway. I The airport authority had issecd a NOTAM in October 1969. stating that there was 9,700 fect available for takeoff on Runway 28R. In accord. ance with this NOTAM, Pan American had taodified the data in their Route Manual and the Re ee Beng OP ee SR SS RS y . . * ' . - cs . shige een geist nats 22 +4 PMs 7 Pt Oe Pa ee ~ charts reflected 9,700 feet available for that runway. The NOTAM did not include the information that runway lights extended only to the displaced threshold and that only 8,896 feet were available for use at night. Additionally, 10 information had been published indicating that the first 200 fect of the load bearing surface of the runway was covered with fill material to stabilive additional area for a future lengthening of this runway. This made that particular portion of Runway 28R unuscable. There was so much confusion as co the actual length available for Runway 28R that the airport engincering personnel, accompanied by the Board's represen. tative, measured it. They found that there was 8,898 feet of lighted runway available. This is the amount of runway that cuuld be used fer a 24-hour, all weather operation. There was an additional 500 to 600 fect of surface prior to the displaced threshold that could be used for daylight takcoff as well as 200 fect of iillcovered ¢ irface. The appearance of this 700 to 800 fect of surface was such that it would make the load bearing capabilities suspect to the pilot of a heavily loaded jet aircraft. Again, the tabulated charts from the Pan Arerican Route Manuals were in error. The error was caused by utilization of the informa. tion that had been provided by the airport authority. ‘There was nothing about the appearance of the approach end of Runway 28R that would cause a pilot to believe its length was other than thet published on the charts. This was a potentially dangerous situation. This accident record revealed several examples of assumptions made by involved personnel that “someone else" verified the accuracy and/or the currency of information being used.

[CONCLUSIONS > FINDINGS] Flight crew did not recompute the required takeoff reference speeds (V speeds) for a 20° Map condition. . Confusion and lack of uniform procedures existed at all levels related to the processing of information to be transmitted to flightcrews through NOTAM, AfRAD, and other means. . Confusion and lack of agreement existed relative to the actual length of the runways at the San Francisco Inter. national Airport. 2.2.2 Probable Cause The National Transportation Safety Board determines that the probable cause of thie accident was the pilot's use of incorrect takeofé reference speeds. ‘This resulted from a series of irregularities involving: (1) the collection and dissemination of airport information; (2) aircraft dispatching; and (3) crew management and aiscipline; which collectively rendered ineffective the air carrier's operational control system. 3. RECOMMENDATIONS As a result of this investigation, the Board recommended on January 3, 1972, that the FAA take the following actions: 4, Review the procedures for the issuance of NOTAM and AIRAD for standardized implementation within the appropriate FAA facilities and modify the. procedures to assure that Information pertinent to “Safety of Flight” is disseminated without delay. “2, Require that V reference speed checks be included on the last checklist used immediately prior to takeoff. “3. Require the installatton of runway distance markers at all civil airports where air carrier aircraft arc authorized to operate. - “4, Require the use of takeoff procedures which will provide the flightcrew with time and distance reference to associate with acceleration to V4 specd. I “§, Requize manufacturers to include in. formation in the Aircraf: Flight Manual concerning the aircraft controllability and performanc? characteristics with the loss of any system that involves flight controls, Consideration should b+ given to incorporating tralaing in such in-flight emtgencics in all approved simulator programs at the earliest possible. date.” On February 24, 1972, the PAA replicd that: 1. They had initiated a study to reevaluate the NOTAM system, Following receipt of comréents from the FAA regions and evaluation ‘by a headquarters (cam, a manual which will consolidate and standardize all information ccucerning NOTAM's will be developed. I 2. They plan to issue an operations bulletin to all their ficld inspectors to ensute that airline training programs emphasize the necessity for flightcrews to assure that takeoff reference speeds include accurate resolution of all pertinent factors prior to initiating a takeoff.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] 2.1 Analysis 2.1.1 Accidem It became apparent, as this invescgation progressed, that the causal factors in this accident were in the operational area. The crew was ptoperly certificated and was qualified to perform its duties in accordance with the existing company and federal regulations applicable to this operation. The flight controllers were certificated and currently qualified to perform the duries required in this operation. The ‘alrcraft performance capability and the physical dimensions of the runway utilized (O1R) were adequate fot the takeoff, if the aircraft had been operated in conformance with the recommended procedures. The actual wind conditions were in excess of the crosswind limits recommended for the utilization of Runway O1R as the preferential runway for takeoff, but these conditions are not considered significant with rsspect to this accident. The invescipation revealed that the “cleatway” for Rurway O1R dic not meet the FAA criteria for a clearway. The regulatory requirements for a clearway were predicated on no cbstruction penetrating a 1.25 percent upward slope beginning at the departure end of the runway. This slope was penetrated by the handrail installed on the access walkway to light platforms of the ALS to Runway 19L. The clearway would alsu have been violated if the channel used by fuel barge had been in use. Failure to meet the clearway criteria was not significant in the first instance where the handrail penetrated the clearway, and academic in the second instance, since there was no barge traffic in the sea lane at the time of the accident. Although these deviations from criteria an rocedure were not found to be causal factors in this accident, they do ustrate the need for sesponsible operating officials to assure by verification rather than by assumption, that operational conditions are in fact as intended, The elevation of the departure end of Runway OIR was 11.07 feet mas. and the handrail at the second light platform 400 feet from the end of this runway, was 16.45 feet m.s.l. However, the floor of the clearway (zhe maximum allowable elevation at this point) was 16.07 feet msl. Finally, the Board found no evidence that a procedure oF method had been established to provide positive control over the fuel barge traffic cited in the Extended Duration Notices to Airmen, section of Part 3 of the Airmap’s Infornration Manual of July 22, 1971. The Board has ocen advised that Pan American no longer uses clearways for takeoff. The pattern of occurrences preceding this accident was initiated when the flight controller lanned and prepared for a heavy jet departure from the longest runway ‘98L) on the airport without ascertaining the status of that runway.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] It also would explain why the Pan American personnel who were responsible for the preparation of the B-747 Route Manual, did tot reflect the 8,400-foot takeoff length for this runway, and why all of the tabulated data for the maximum gross weight limits tor that runway were in error. The FAA tower coordinator on duty at the time of this accident was also an assistant tower chief, As an assistant tower chief, he was aware of the noise complaint situation that occurred almost every time 28h was used as the departure runway. He was conversant with the recommended crosswind limits of 80° and 15 knots which applicd to the preferential use of Runway OIR for noise abatement nurposes, burt he favored the use of Runway 61 for departure any time the conditions allowed. He was aware that the crosswind at the time of the accident was exceeding the recommended limits in both 20 direction and velucity. He specifically directed that PA 845 be given the wind conditions just pzior to the takeoff clearance. It is now apparent that he was assuring that the pilots of departing aircraft were informed of the wind conditions. He did not change the runway or recommend the use of Runway 28R, The Board believes that, where recommended limits on runway use have been established by the Federal Aviation Administration, the representatives of the Administrator shuuld adhere io those limits as closely as possible. While the crosswind was not a critical factor in this case, no operation should be condoned that, as 2 matter of expediency, jeopardizes or degrades the established margins of safezy. If the only reason for exceeding established operating limits {{recommended or mandatory) is expediency, then the operation should be discontinued rather than jeopardize the crew, passengers, or persons on the ground. The flight controller responsible for the flight did not receive the initial radio call from PA 845 and was unaware of the request fur information about 28R. He simply rever:2d to his original alternate planning consideration of OFR with clearway for departure. Since Runway 28R was favored by the wind. the desirability of its use for the takcoff was examined by the Board. The actual length of Runway 28R was not known to those required to know and to publish it. Two different lengths were given by representatives of the airport authority and a third length was listed in the airport diagram attached to the FAA Airport Master Record (FAA FORM 5010-1) dated April 1, 1971.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] The Board has ocen advised that Pan American no longer uses clearways for takeoff. The pattern of occurrences preceding this accident was initiated when the flight controller lanned and prepared for a heavy jet departure from the longest runway ‘98L) on the airport without ascertaining the status of that runway. This runway was routinely available aud a closure of the runway wes the exception to the rule rather than the normal circumstance. The flight controller assumed that all conditions were routine, and did not check the available information sources. He was subjected to one © the most insidious hazards facing any routine operation, that of being lulled into a sense © complaceny. This condition would exist because previous checks he had made for unusual conditions were routinely unfruitful and he had no reason to suspect that this day would be different. The flight controllers availed themselves of the telephone briefing system only when something unusual occuszed. If they relied on this 17 procedure as a safeguard, then the chief flight controller, who listened to the ATIS broadcasts on an hourly routine echedule should have used the telephone service as soon as the information about Runway 28L being closed was omitted from the “Whiskey” ATIS issued at 1230. it was incumbent upon the supervisor of the shift to ascure himself that the change in runway status was accurate and that all flight controllers working for him had that information. The procedures for communicating required information were lax within the dispatch office at San Francisco. There was no formal procedure for briefing the flight controller on the conditions and circumstances existing, OF expected to exist, Over the part of the airline's system for which this office was responsible. Since the flight controller responsible for PA 845 was not at his duty post until about 1145, he hac no opportunity to learn from any ATIS broadcast of the closure of Runway 28L prior ‘0 the actual flight release planning. The contra: dictory statements made by the relicf controller, concerning the information contained in the ATIS broadcast from 1100 to 1200. would indicate that an accurate briefing could not have been given by him and the sentot flight control: ler (supervisor) stated that he did not brief the subject controller upon his return from taking the company physical examination. The Board believes that the requirement for a proper bricfing Is a supervisory responsibility that cannat be delegated or omitred. Before the crew taxied the aircraft out for cakeoff, they learned that Runway 28L was closzd. They called Pan American Dispatch via an ARINC tadio frequency and requested them to check the feasibility and lepality of using Runway 28R for departure. The fight controller responsible for the flight did not receive the call but he was notified that the crew of PA 845 was having a problem and was requesting a change © runways duc to the closure of Runway 28L.

[CONCLUSIONS > FINDINGS] The aircraft's weight, center of gravity, and load distribution was within the established limits. There was sufficient runway length available for a successful takeoff from Runway 01R, however, the runway length did not mect the FAA criteria for a takeoff under the existing conditicns. The FAA tower controller utilized a preferential ‘akeoff runway with existing wind conditions in excess of the recommended crosswind limitations. The Pan Amertean flight controliers were certificated and qualified for their assigned dutics. _ The takeoff gross weight computation by the flight controller was based on the use of Runway 28L for the departure of PA 845. . The Pan American flight controller did not checl. the existing and forecast airport conditions prior to planning for the departure of PA 845. . Runway 28L was closed for repair during the planning for PA 845's departure and was forecast to be closed until after PA 845's scheduled departure time. . The closure of Runway 28L and the first 1,006 feet of O01R was not included in the appropriate NOTAM. . There was a restriction against B-747 and other specific types of alrceaft utilizing full takeoff thrust prior to reaching the displaced threshold on Runway OIR. There was 8,400 feet of runway olus a clearway authorized for B-747 takcoffs from Runaway O1R on the date of this accident. . The ALS structure for Runway 19L penctration of the clearway and the entry of barges across the takeoff zone ‘for Runway O1R negated the availability of a clearway for a 01R departure. . The tower controller assumed that the flight controller wes familiar with the B-747 restriction ¢n the use of O1R when the flight controller called the tower concerning the closure informa. tlon carried on ATIS information “XRAY” and responacd to the flight controller accordingly. . The Pan American B-747 Route Manual computations were based upon the assumption that Runway O1R had 9,500 feet available for takeoff. . PA 845 was configured to 10° of flaps and reference speeds (V4-156 knots, Vp-164 knots, V9-171 knots) were set pie pee cempciettll SSL Many aera Te on the airspecd indicator bugs for takeoff on Runway 28L before the cr-w learned it was closed. . PA 845 was configured to a 20° flap setting for takeoff from Runway O1R while in the runup arca. . The takeoff reference speeds shown in the Pan American B-747 Aircraft Operating Manual for PA 845’s gross weight and a 20° flap setting were: V1-149 knots, Vp-157 knots, V2-162 knots . The.

[ANALYSIS] Although the weather conditions were sufficient for the approach to be made safely, the low overcast cloud layer and heavy rain and fog encountered by flight 554 during its approach to runway 13 degraded visual cues that the captain might otherwise have used to gauge the airplane’s rate of descent/descent path during the visual portion of the approach. As discussed in section 1.18.3, according to the FAA AIM, visual illusions that might lead a pilot to perceive that the airplane is higher or more distant from the runway than it is during an approach can result from the following conditions: • an absence of ground features [as when landing over water] • rain on the windscreen • atmospheric haze/fog • terrain with few lights to provide height cues The Safety Board notes that all of these conditions were present when the pilots of Delta flight 554 descended out of the overcast cloud layer and the captain transitioned to visual conditions. According to FAA and medical publications55 on the subject of visual illusions, these conditions could result in improper perception of altitude and descent path; specifically, a pilot might perceive the altitude to be higher than the airplane’s actual altitude, especially during periods of reduced visibility, when other visual cues are not available. Further, the runway 13 edge lights were spaced irregularly—most of the lights were spaced at intervals less than the maximum interval of 200 feet set forth in AC 150/534024—and the departure end of runway 13 was obscured by rain and fog, so the pilots were visually presented with a foreshortened runway. Pilots who are accustomed to operating into airports at which runway lights are spaced at consistent 200-foot intervals might perceive their distance and angle to the runway differently when presented with runway lights spaced at shorter, irregular intervals. 55 FAA AIM and Fundamentals of Aviation Medicine. 59 The Safety Board concludes that the irregular and shortened runway edge light spacing and degraded weather conditions can result in a pilot making an unnecessarily rapid descent and possibly descending too soon, especially in the absence of other visual references or cues. Therefore, the Safety Board believes that the FAA should identify Part 139 airports that have irregular runway light spacing, evaluate the potential hazards of such irregular spacing, and determine if standardizing runway light spacing is warranted. Although the airport and weather conditions that existed at the time of the accident combined with the irregular (and shortened) spacing of the runway lights presented a potential challenge for any pilot landing on runway 13, other airplanes used the ILS DME approach to runway 13 around the time of the accident and landed without incident. In an effort to understand why the captain of Delta flight 554 was unable to land safely, the Safety Board analyzed the effect that his use of MV contact lenses had on his vision under those conditions. Individuals with normal binocular vision use both binocular and monocular cues for depth perception.

[ANALYSIS] However, the captain was using monovision (MV) contact lenses, which were not approved by the FAA for use by pilots while flying. The captain indicated in postaccident interviews that he was not aware that MV contact lenses were not approved for use while flying. The flight attendants had completed Delta’s FAA-approved flight attendant training program. The airplane was properly certificated, and there was no evidence that airplane maintenance was a factor in the accident. A review of ATC procedures revealed that the controllers followed proper air traffic separation rules, and air traffic separation was assured during flight 554’s approach to the runway. In addition, ATC provided the pilots with timely weather (rain and visibility) information during their approach to runway 13. No ATC factors contributed to the cause of the accident. Although the pilots did not receive several pieces of weather information, Delta Air Lines provided the pilots with sufficient preflight, en route, and arrival weather information to allow them to conduct the flight safely; however, because of rapidly changing surface conditions, the conditions they encountered differed from what was forecast. Although the pilots encountered a low cloud ceiling and degraded visibility, the ceiling and visibility were still above the minimums required for the ILS approach. The Safety Board concludes that although the weather conditions encountered by the pilots during the approach differed from the forecast conditions, these conditions should not have affected the pilots’ ability to conduct a safe approach and landing. The pilots reported that they did not encounter much turbulence after they descended through 3,000 feet on the approach, and the recorded wind speeds were less than expected. FDR data from flight 554 and four other airplanes that made the approach to runway 13 between 9 minutes before and 1 minute after the accident disclosed no evidence of significant windshear. Additionally, no pilot reports of windshear, LLWAS windshear alerts, or wind gusts were recorded during the 50 minutes preceding the accident. Therefore, the Safety Board concludes that Delta flight 554 did not encounter windshear during its approach to runway 13 at LaGuardia. The pilots performed the instrument approach to runway 13 in low clouds, moderate-to-heavy rain, fog, and in limited light conditions. 56 Although the pilots did not observe the VASI lights during the approach to runway 13, postaccident examination revealed that the VASI light system was capable of normal operation at the time of the accident. Although it is likely that the runway 13 VASI light system was operating normally when the accident occurred, with the reduced visibility that existed at the time of the accident, the pilots would not have been able to utilize descent path guidance from the VASI light system until late in the visual phase of the approach, and then only if they sought such guidance. 2.2 The Accident Scenario 2.2.1 The Approach and Descent The pilots reported that the departure, en route, and initial approach portions of the flight were routine.

[CONCLUSIONS > FINDINGS] The current requirements for special airport pilot qualifications might not be sufficient to ensure that pilots who are so qualified have been exposed to the runways and/or approaches at those airports that make the airport “special.” 69 22. The flightcrew coordination appeared adequate, and the decision to evacuate the airplane was appropriate and timely. 23. The flight attendant in charge, who began shouting evacuation commands within 2 seconds of the evacuation order, reacted to the evacuation command promptly and assertively, in accordance with Delta’s flight attendant manuals and training. 24. The two aft flight attendants did not react promptly or demonstrate assertive leadership, as specified in Delta’s flight attendant manuals and training. 25. The quality of the crew resource management was not a factor in this accident. 26. The atypical installation and use of runway visual range transmissometer equipment at LaGuardia did not adversely affect the validity of the runway visual range values reported at the time of the accident. 27. The low level windshear alert system equipment anomalies were not a factor in this accident. 70 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the inability of the captain, because of his use of monovision contact lenses, to overcome his misperception of the airplane’s position relative to the runway during the visual portion of the approach. This misperception occurred because of visual illusions produced by the approach over water in limited light conditions, the absence of visible ground features, the rain and fog, and the irregular spacing of the runway lights. Contributing to the accident was the lack of instantaneous vertical speed information available to the pilot not flying, and the incomplete guidance available to optometrists, aviation medical examiners, and pilots regarding the prescription of unapproved monovision contact lenses for use by pilots. 71

[ANALYSIS] In addition, the information provided in AC 121.445-1D’s remarks section is general, and does not provide operators with detailed information as to the justification for special airport designation, nor does it describe specific approaches, runways, hazards, or obstacles. The Safety Board concludes that the FAA’s current guidance on special airports contained in AC 121.445-1D is not sufficiently specific about criteria and procedures for designation of special airports; therefore, the FAA’s current guidance might not always be useful to air carriers operating in and out of (existing or potential) special airports. The Safety Board is aware that the FAA met with industry representatives in October 1996 to develop a list of factors—based on accident histories, human performance concerns, runway anomalies, etc.—to use in determining criteria for classification of special airports. However, the Safety Board is concerned that the FAA has apparently not made any progress in developing such criteria since that meeting. Therefore, the Safety Board believes that the FAA should expedite the development and publication of specific criteria and conditions for the classification of special airports; the resultant publication should include specific remarks detailing the reason(s) an airport is determined to be a special airport, and procedures for adding and removing airports from special airport classification. The Safety Board is also concerned that if an airport is designated “special” because of a specific approach or runway configuration (i.e., the ILS DME approach to runway 13 at LaGuardia) a pilot who satisfies the special pilot qualification requirements by landing and departing on a different runway at that airport might not have appropriate familiarization with the special features of that specific approach or runway configuration and therefore might not adequately satisfy the intent of the special airport regulation. The Safety Board concludes that the present requirements for special airport pilot qualifications might not be sufficient to ensure that pilots who are so qualified have been exposed to the runways and/or approaches at those airports that make the airport “special.” Thus, the Safety Board believes that the FAA should develop criteria for special runways and/or special approaches giving consideration to the circumstances of this accident and any unique characteristics and special conditions at airports (such as those that exist for the approaches to runways 31 and 13 at LaGuardia) and include detailed pilot qualification requirements for designated special runways or approaches. Also, the Safety Board believes that once criteria for designating special airports and special runways and/or special approaches have been developed, the FAA should evaluate all airports against that criteria and update its special airport publications accordingly. 2.7 Flight and Cabin Crew Evacuation Actions The Safety Board considers that in general, the crewmembers’ responses after the airplane came to a stop were commensurate with the circumstances of this accident. First, the crewmembers assessed the condition of the airplane and reviewed their options; then, when the captain was informed that there was a smell of jet fuel fumes in the passenger cabin, he promptly commanded an emergency evacuation.

[ANALYSIS] Analysis 44 Aircraft Accident Report After the airplane’s third touchdown, the data showed three simultaneous events indicating that the pitch uncoupling mechanism had uncoupled. First, the control column positions deviated from their normally recorded deviations. Second, the left and right elevators began to move independently of each other. Last, the FDR pitch uncoupling parameter showed “warn.” None of these events were present before the third touchdown. Therefore, the Safety Board concludes that the pitch control uncoupling mechanism uncoupled when the airplane touched down for the third time; as a result, the pitch uncoupling would not have prevented the flight crew from controlling or safely landing the airplane. 2.7 Failure of the Left Main Landing Gear CVR, FDR, and wreckage and impact information indicated that the airplane’s third touchdown severely damaged the left MLG, which caused it to fracture circumferentially at its vertical trunnion leg just below the actuator attach point. Examination of the fracture surfaces revealed evidence of overload failure. No evidence of preexisting defects was found. The ATR-72 landing gear and associated structure were designed to absorb energy equivalent to a maximum airplane descent rate of 10 fps when landing at the airplane’s maximum design landing weight (consistent with the landing design limits imposed by 14 CFR 25.473 to 25.487). However, the Safety Board’s airplane performance study determined that, when the airplane touched down for the third time, the left MLG touched down at a descent rate of about 19 to 32 fps. Therefore, the Safety Board concludes that, when the airplane touched down for the third time, the vertical forces on the left MLG exceeded those that the gear was designed to withstand, and these excessive forces resulted in overload failure. 45 Aircraft Accident Report 3. Conclusions 3.1 Findings 1. The captain was properly certificated and qualified under Federal regulations. No evidence indicated any preexisting medical or physical conditions that might have adversely affected his performance during the accident flight. The first officer held a current Federal Aviation Administration airman medical certificate at the time of the accident; however, he failed to provide information about his medical condition (anxiety) or his use of the prescription drug alprazolam when he applied for the certificate. 2. The airplane was properly certificated, equipped, and maintained in accordance with Federal regulations and approved company procedures. The airplane was loaded in accordance with approved company weight and balance procedures. The weight and balance of the airplane were within limits during all phases of the flight. 3. Winds were within the airplane’s performance capabilities and did not adversely affect the flight crew’s ability to maneuver the airplane during the approach and landing as significant aircraft control authority remained. 4. The emergency response was timely and appropriate.

[CONCLUSIONS > FINDINGS] The airplane was loaded in accordance with approved company weight and balance procedures. The weight and balance of the airplane were within limits during all phases of the flight. 3. Winds were within the airplane’s performance capabilities and did not adversely affect the flight crew’s ability to maneuver the airplane during the approach and landing as significant aircraft control authority remained. 4. The emergency response was timely and appropriate. The passengers and crewmembers were safely evacuated from the airplane. 5. At some point during the accident sequence, the captain cockpit seat failed when it was subjected to vertical loads that exceeded those required for certification. 6. The flight crewmembers did not account for winds when calculating the minimum approach airspeed, and, as a result, they were not in compliance with Executive Airlines’ approach airspeed procedures. 7. Given the relative positions of the accident airplane and the preceding Boeing 727, the runway configuration, and the existing winds, wake turbulence was not a factor in this accident. 8. The captain did not properly follow Executive Airlines’ before landing procedures. 9. The flight crew could have completed a successful landing after the initial touchdown. 10. After each bounce of the airplane on the runway, the captain did not make appropriate pitch and power corrections or execute a go-around, both of which were causal to the accident. Conclusions 46 Aircraft Accident Report 11. The captain demonstrated poor cockpit oversight and piloting techniques before and during the accident sequence. 12. Written company guidance on bounced landing recovery techniques would have increased the possibility that the captain could have recovered from the bounced landings or handled the airplane more appropriately by executing a go-around. 13. The performance of air carrier pilots would be improved if additional guidance and training in bounced landing recovery techniques were available. 14. The aileron flight control surface position sensors installed on airplanes in accordance with Supplemental Type Certificate No. ST01310NY are unreliable, and flight data recorder functional checks every 6 months could ensure the timely identification and correction of potentiometer malfunctions and ensure that accurate flight control data are available for accident and incident investigations. 15. Because the first officer started getting treatment for anxiety in July 2001, he should have reported this information on his last three Federal Aviation Administration airman medical certificate applications. 16. Not enough evidence was available to determine whether or to what extent the first officer’s medical condition and prescription drug use contributed to the accident. 17. The pitch control uncoupling mechanism uncoupled when the airplane touched down for the third time; as a result, the pitch uncoupling would not have prevented the flight crew from controlling or safely landing the airplane. 18. When the airplane touched down for the third time, the vertical forces on the left main landing gear exceeded those that the gear was designed to withstand, and these excessive forces resulted in overload failure. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the captain’s failure to execute proper techniques to recover from the bounced landings and his subsequent failure to execute a go-around. 47 Aircraft Accident Report

[ANALYSIS] Wreckage and impact information and CVR and FDR data indicated that the captain’s actions after the initial touchdown resulted in the airplane bouncing on the runway twice. The captain’s inputs made it less likely that he could 60 Executive Airlines’ AOM states that the condition levers can remain at 86 percent for landing the ATR-72 and that maximum rpm could be set at the pilot’s discretion. After the accident, the company standardized its procedures, requiring that the condition levers on all of its ATR airplanes be set to 100 percent before landing. 61 Executive Airlines’ AOM states that the nonflying pilot should set the power management selector to TO before landing. 62 As noted previously, this flight was the first officer’s first since he completed IOE; therefore, it was appropriate for the captain to instruct the first officer during the flight. Analysis 40 Aircraft Accident Report recover from the two bounces and safely land the airplane; as a result, he should have executed a go-around. The Safety Board concludes that, after each bounce of the airplane on the runway, the captain did not make appropriate pitch and power corrections or execute a go-around, both of which were causal to the accident. On the basis of the evidence presented in this section, the Safety Board concludes that the captain demonstrated poor cockpit oversight and piloting techniques before and during the accident sequence. 2.3 Bounced Landing Recovery Guidance and Training Postaccident interviews with the first officer, two company check airmen, and three company simulator instructors revealed that Executive Airlines did not have standardized guidance regarding bounced landing recovery. For example, one of the check airmen and one of the simulator instructors stated that, if the airplane bounced, they would execute a go-around. The other three company personnel indicated that, if possible, they would try to correct the bounce and land and that, if not possible, they would execute a go-around. Further, Executive Airlines’ manager of training and standards stated that, before the accident, the company did not teach its pilots bounced landing recovery techniques. The manager also stated that he would not want to conduct bounced landing recovery training in the simulator because it was very difficult to demonstrate. However, he stated that, after the accident, the president and the vice president of operations asked him to look into the feasibility of conducting bounced landing recovery flight training and incorporating bounced landing recovery techniques in company manuals. The Safety Board concludes that written company guidance on bounced landing recovery techniques would have increased the possibility that the captain could have recovered from the bounced landings or handled the airplane more appropriately by executing a go-around. As noted previously, on September 25, 2004, Executive Airlines incorporated bounced landing recovery techniques in its AOM. In its final report on the July 31, 1997, Federal Express flight 14 landing accident,63 the Safety Board concluded that the captain’s overcontrol of the elevator during the landing and his failure to execute a go-around from a destabilized flare were causal to the accident.

[ANALYSIS] At 1444:57, the captain stated that he was setting the condition levers to 86 percent.60 At 1445:02, the first officer called for the before landing checklist. About 1 minute later, the captain stated that he was going to move the power management selector from TO to CLB “just for now” to prevent the condition levers from automatically advancing to 100 percent. However, the CVR did not record the captain repositioning the power management selector back to TO.61 Postaccident documentation of the cockpit switch positions by the Safety Board’s Operations Group confirmed that the power management selector was set to CLB. Therefore, the Safety Board concludes that the captain did not properly follow Executive Airlines’ before landing procedures. At 1446:39, the SJU local controller cleared flight 5401 to land, and the first officer turned the airplane left toward the runway and monitored the VASI for glideslope guidance for the rest of the approach. At 1449:07, the captain told the first officer to keep the airplane’s nose down or to increase power (to maintain airspeed) because he was concerned that the airplane was going to balloon. About 2 seconds later, he instructed the first officer to get the airplane’s nose up, and, at 1449:28, he told the first officer to “power in a little bit.”62 About 2 seconds later, the airplane crossed the runway threshold at an airspeed of about 110 KIAS, which was almost 15 knots more than the Vref set on both pilots’ airspeed indicators (96 KIAS) and 9 knots more than what should have been set. After crossing the runway threshold, the captain again told the first officer to “power in a little bit” and not to pull the airplane’s nose up. CVR and FDR data indicated that the airplane touched down on the runway the first time about 1449:41 and then skipped and became airborne, reaching an altitude of about 4 feet. Only minor flight control inputs and/or slight power adjustments would most likely have been necessary to regain the proper landing attitude and settle the airplane back on the runway. Therefore, the Safety Board concludes that the flight crew could have completed a successful landing after the initial touchdown. After the initial touchdown, the captain took control of the airplane, most likely because of the first officer’s inexperience. FDR data indicated that he then made several abrupt changes in pitch and power. Wreckage and impact information and CVR and FDR data indicated that the captain’s actions after the initial touchdown resulted in the airplane bouncing on the runway twice. The captain’s inputs made it less likely that he could 60 Executive Airlines’ AOM states that the condition levers can remain at 86 percent for landing the ATR-72 and that maximum rpm could be set at the pilot’s discretion.

[ANALYSIS] As noted previously, on September 25, 2004, Executive Airlines incorporated bounced landing recovery techniques in its AOM. In its final report on the July 31, 1997, Federal Express flight 14 landing accident,63 the Safety Board concluded that the captain’s overcontrol of the elevator during the landing and his failure to execute a go-around from a destabilized flare were causal to the accident. As a result, the Board issued Safety Recommendation A-00-93 to the FAA, which required, in part, that a syllabus for simulator training be developed that addressed how to recover from unstabilized landing flares, including techniques for avoiding and recovering from overcontrol in pitch before touchdown and techniques for avoiding overcontrol and premature derotation during a bounced landing. In a May 15, 2002, letter, the FAA stated that an industry taskforce had been convened and that the taskforce had produced several significant training materials, including an ALAR 63 NTSB/AAR-00/02. Analysis 41 Aircraft Accident Report Training Guide, to address the safety recommendation. On October 22, 2002, the Board classified Safety Recommendation A-00-93 “Closed—Acceptable Action.” The training materials produced in response to Safety Recommendation A-00-93 do not specifically address bounced landing recovery techniques. Further, an informal Safety Board survey of six airlines, an airplane manufacturer, and a pilot training facility revealed that only some of the companies included bounced landing recovery techniques in their flight manuals and discussed these techniques during training. The Board is concerned that the lack of guidance on bounced landing recovery techniques could contribute to similar landing accidents. The Safety Board concludes that the performance of air carrier pilots would be improved if additional guidance and training in bounced landing recovery techniques were available. Therefore, the Safety Board believes that the FAA should require all 14 CFR Part 121 and 135 air carriers to incorporate bounced landing recovery techniques in their flight manuals and to teach these techniques during initial and recurrent training. 2.4 Quality of Data Provided by Flight Data Recorder Potentiometer Sensors The Safety Board determined that the left aileron surface position data recorded by the accident airplane’s FDR were invalid even though the accident airplane had been modified on August 7, 2001, with the position sensors and associated hardware required by STC No. ST01310NY. Executive Airlines stated that the first and last FDR parameter readout since the installation of the sensors occurred on January 3, 2003, about 1 year and 5 months after the installation and 1 year and 4 months before the accident. Executive Airlines stated that it had replaced 47 aileron surface position sensors in the last 3.5 years. However, the company indicated that, because the sensors are not tracked, the times from installation to failure could not be determined. During its investigation of the Atlantic Southeast Airlines flight 529 accident,64 the Safety Board determined that the EMB-120’s two flight control position sensors had malfunctioned, preventing the required data from being accurately recorded, and that the lack of data hindered the investigation.

[ANALYSIS > THE ACCIDENT] It is recommended that the sterile cockpit light be turned on at 10,000 feet.” The "In Range" checklist contains the following: "Note: Captain will ascertain proper time to turn sterile cockpit light on." 2. ANALYSIS 2.1 The Accident The investigation revealed that the flightcrew was properly certificated and qualified to conduct the flight. The airsraft was properly certificated, equipped, and ~20- maintained. The landing gross weight was within limits for the reported winds and the 30° flap setting. The overtoud condition imposed on the right landing gear was caused by two conditions: the upslope of the area where the touchdown was made and the descent rate of the aircraft at touchdown. The investigation revealed that the shear load imparted to the landing gear as a result of the upsloping terrain was 824 ft/min (13.74 [t/sec), which would have been below the design strength if tne aircraft had beer on a level runway. Similarly, the calculated vertical descent rate (600 to 750 ft/min) would have imparted a shear load to the landing gear well below the design strength for a touchdown on a level runway. The combination of the two forees, howevcr, exceeded the design strength of the gear. Also, the right main landing gear sustained the full force of the impact without the left main landing gear sharing the load of a simultaneous contact. Therefore, the combination of the upslope at the touchdown point and the vertical descent of the aircraft caused the right main landing gear to separate. The Safety Board's analysis of the evidence in this accident focused on the reasons why the aireraft landed short of the rur.:‘ay. The investigation revealed no mechanical or meteorological reason which coud have caused the short landing. Examination of the wreckaze and a kinematic analysis of ithe dynamics of the touchdown revealed that the design strength of the right main lanc’ .¢ gear structure was exceeded by the forces of the impact. The right main landing gear separated as designed, precluding worse damage to the wing and fuselage structure and preventing a serious fuel spill at impact. The events subsequent to the initial touchdown were incidental only to the survival aspects of the accident. It is apparent from the statements of the four flightdeck occupants and crom the CVR and FDR information that the landing pattern at Yap was flown low and flat, which was not the standard prescribed procedure. Nevertheless, all four flightdeck oceupants believed that the aircraft was going to make a safe landing until tie aircraft was about 50 feet above the runway and the captain reduced the thrust to idle. Although the first officer, second officer, and the mechanic were concerned about the final approach being low, they apparently believed the aircraft would land on the runway until the power was reduced. The captain stated that he still believed that the aircraft would Jand on the runway, although closer to the threshold than he had planned. Airspeed was maintained at or near reference speed until the point where power was reduced about 50 feet above the runway. At that point, the descent rate increased rapidly when the thrust was reduced to idle.

[ANALYSIS > THE ACCIDENT] If he had turned on the final approach at the same distance but at the proper altitude of 500 feet, he would have been on a normal 3° approach slope angle to the 1,000-foot aim point. However, the low base~-leg aititude and turn to the final approach required a flat approach slope angle of about 1.5° and a low rate of descent. He probably flew the a»proach in this manner to attempt a short field-type landing. Because he failed to establish a proper glidepath, his sight picture of the runway, as compared to a standard pattern, would have been abnormal, and more thrust would have been required to hold the lower-than-normal descent rate. This type of dragged-in, flat approach places an aircraft in a difficult situation with respect to windshear, downdrafts, or loss of thrust. Because the margins for error are much less in this type of approach, the FAA and airline companies prescribe standard stabilized approach procedures for jet transport category aircraft. A standard flight pattern procedure by the captain was all the more important in this ease because this was his first unsupervised landing at Yap since he resumed flying a B-72" aireraft. His recent requalification in the B-727 and limited familiarity with Yap should have alerted him to use the preseribed procedures. If he had, he would have had a greater margin for error. If he had reduced the throttles to idle at 50 feet over the runway surface during a prescribed approach, a hard landing probably would have resulted, but it is not likely the aircraft would have been damaged. Tre transition to a landing attitude begun at 5(}} feet from a normal 3° approach slope angle and the prescribed smooth thrust reduction will generally result in a normal lending, whereas a dragged-in, flat approach requires excess power. 2.2 Training Aspects The Safety Board believes that the captain's premature reduction of thrust on this final approach may have resulted from a habit pattern developed during nis previous experience in landing the DC-30. Specifically, the DC-10 has mass/energy and aerodynamic characteristics which produce a greater tendency to float in ground effect than does the B-727. Further, the DC-10 does not necessarily require comparatively es much thrust carried until at or near touchdown as does the B-727. Thus, the captain's prior experience in landing the DC-10 could have contributed to the development of & thrust reduction habit pattern which, although appropriate to the DC-10, was not appropriate for the B~727, especially during a low, flat approach in the B-727. The captain certainly should have been aware of the aircraft differences from his training; however, he did have a iong delay from his last B-727 training flight to his first line fight (61 days). He also returned to flying the DC-10 before his 8-727 line flying. This training sequence and time factor does occur in routine airline operations, especially following a reduction-in-force or other schedule changes.

[ANALYSIS > THE ACCIDENT] The captain stated that he still believed that the aircraft would Jand on the runway, although closer to the threshold than he had planned. Airspeed was maintained at or near reference speed until the point where power was reduced about 50 feet above the runway. At that point, the descent rate increased rapidly when the thrust was reduced to idle. Even thvugh the control yoke was probably pulled aft in an alieimpt to maintain the approach path, without power the airspeed decrcased rapidly and the deseent rate increased rapidly because the aircraft had insufficient thrust in relation to drag to reach the runway. Therefore, the aircraft landed short of the runway because the captain prematurely reduced the thrust. There are several reasons why the captain arrived at a point in this approach where he mistakenly reduced thrust and landed short. Of these reasons, the one of major concern to the Safety Board was the manner in which the approach was flown. The Safety Board believes that the captain's failure to fly a standard, approved pattern directly contributed to the final outeome. It was apparent from the captain's statements that he was ecneerned about the short runway, and that he intended to touch down before the company-prescribed touchdown point of 1,000 feet. The captain's training in both the DC-10 and B-727 aircraft and flight manual procedures emphasized the need to plan a pattern for a touchdown aim point of 1,000 feet beyond the threshold of the runway. Admittedly, the length of the runway at Yap (4,820 feet) is comparatively short; however, Bt pi ee Spare ite oe PNY tS pte RS Bale ABI Fm be tee ee the stopping procedures anc certification data for the aircraft insure a safe landing if recommended pattern procedures are followed. The Safety Board believes: that the captain was ignoring these criteria and was concerned about the short length of the runway; therefore, he planned to land about 300 feet rather than 1,000 feet beyond the runway threshold. The approach to Yap was not typical of the type previously flown by the capiain. The fly-by procedure to cheek the runway placed the aircraft in an abnormal position on the downwind leg of the pattern. Once the fly~by was completed, however, the captain was required to establish a normal bese leg and final approach. In this case, the captain did not regain the proper altitude for a normal base leg; instead he turned for the final approach about 1.5 miles from the runway at only 250 feet above the runway slevation instead of being stabilized on the final at 500 feet as recommended in the approved flight manual. If he had turned on the final approach at the same distance but at the proper altitude of 500 feet, he would have been on a normal 3° approach slope angle to the 1,000-foot aim point. However, the low base~-leg aititude and turn to the final approach required a flat approach slope angle of about 1.5° and a low rate of descent.

[ANALYSIS > THE ACCIDENT] The fact that the aft arstair exit was not opened was nearly catastrophic because one flight attendant and som. passengers were almost trapped in that area. It coule not be determined if the pneumatic ~%4- emergency blow-down system would have forced the exit open; however, the fact that the fiight attendant did not know how to actuate the emergency system is a serious concern. Her repeated attempts to open the exit using the normal systein delayed her evacuation to a point where she was nearly trapped by the smoke and fire. CONCLUSIONS - Findings The flighterew was properly certificated and qualified to conduct the flight. The aircraft was properly certificated and maintained in accordance with preseribed procedures. The aireraft touched down on the right main landing gear 13 feet short of the approach end of the landing runway. The right main landing geur separated at initial ground contact. The area of initial touchdown of the right main landing gear tires sloped upward about 4.07°. The combined forces of the excessive sink rate and an unsloping touchdown point exceeded the design strength of the right main landing gear. The captain flew a flat, dragged-in final approach with about a 1.5° glide slope which required excess thrust. The first and second officers and the mechanic in the cockpit jumpseat were concerned about the approach being low. The captain reduced the throttles to idle 50 feet above the runway elevation, and short of the runway threshold. The landing was the first unsupervised landing at Yap for the captain. The captain had been flying DC-10 aircraft as captain for about 3 1/2 years prior to November 1980. The captain had not landed a B-727 aircraft for 61 days before the date of the accident. He made one tending, at Saipan, on the day of the accident. Fire erupted around the damaged right wing area as the aircraft came to a stop. The crash forces were not sufficient to cause serious impact injuries to the occupants. q ia *, ; -25~ Thee acuation was completed in about 55 seconds. The flight attendants were not aware of how to open the aft airstair exit door using the emergency system. Immediately following the accident investigation, the airline implemented new training techniques to include "hands-on" training on the aft airstair exit emergency opening system. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the captain's premature reduction of thrust in combination with flying a shallow approach slope angle to an improper touchdown aim point. These actions resulted in a high rate of descent and a touchdown cn upward sloping terrain short of the runway threshold, whieh generated loads that exceeded the design strength and failed the right landing gear. Contributing to the accident were the captain's lack of recent experience in the B-727 aircraft and a transfer of his DC~10 aircraft landing habits and techniques to the operation of the B-727 aircraft.

[ANALYSIS] Finally, this theory is refuted by the lack of immediate concern shown by the crew. Only a few notuncommon remarks concerning the hard touchiown were made in the cockpit. It was not until slightly before the second touchdown when the voice record began to show a sense of {{mpending emergency in the tone cf voice and the comments made ky the crew. Based upon the preceding evidence, the Board concludes that the initial touchdown was not of a destructively hard nature. The next possibility explored was that some malfunction of $$the aircraft caused this accident. Malfunctions which might possibly have been involved include loss of thrust, control system malfunction, landing gear malfunction, and pilot seat failure. The first two ct ee at GOREN SEE 23 te cine ora) malfunctions may be dismissed summarily since the crew did not report any problens in these areas and since no evidence of such malfunctions was observed in the Board's examination of the wreckage or of the flight recorder data. The possibility that a malfunction of the right main landing geax (RMLG) caused this accident precipitated an extensive study of that system. This study produced no evieence that any preimpact malfunction existed in the landing gear or its attach structures; rather, it demonstrated that the parts examined were sound, and that all fractures were caused by overloczrds applied to the RMLG. The pxobable failuice sequence advanced by the Boeing Company seems reasonable to the Board. The second touchdown, which was described as the hardest of the three, overstressed the year attach structure and the failure was jnitiated at that time. Passengers, it should be noted, described grinding sounds, and some thought that something on the RMLG broke at that time. The aircraft then apparently became airborne, after the FS 940 frame began to separate from the fuselage skin, but before the separation was complete. The pieces of fasteners reportedly found about 2,000 feet down the runway are consistent with this theory. The third touchdown then completed the failure of the gear attach structure. . This final disruption of its attach structure then allowed the landing gear to displaced upward until the right wing tip and the trailing edge of the outer wing flaps began to drag on the runway. Having thus determined that the environmental and machine elements were not likely causal factors of this accident, we now turn to the man and man/machine aspects of the operation, A study of the probable sequence of events which occurred during the approach and landing,and the factors which influenced those events, will show more clearly the involvement of the crew in the causal areas of the accident. In this respect, certain aspects of the approach of Flight 505 seem noteworthy. « 27 - One such aspect is the somewhat reversed student ~ instructor relationship which developed in the cockpit @uring the approach. ‘This relationsnip was evident in the decision of the captain to experience the rasponse of the aircraft at slower speeds while on final approach. The factors influencing this decision will be discussed later in this report.

[ANALYSIS] Such a maneuver could also explain the high angle of attack the aircraft assumed upon becoming airborne again. The comments of the crew indicate that a wind shear might have caused the short touchdown, or that the aircraft encountered a downdraft just after it crossed the chreshold. However, no evidence that the flight encountered either phenomenon was found. The surface winds reported by the tower never exceeded 10 knots. Although the firet officer referred to gusty winds during the approach segment. of the Flight, ‘the Board must conclude that these conditions did not exist at the runway threshold; indeed, he did amend his comment. “windy gusty" with the comment “Aw I mean windy out over the ocean there. . . .*# In addition to the possibility that it wa caused by a downdaft or gust, the short touchdown could be xplained by an faliure of the pilot to flare tha airccaft before touchdown. If the aircraft were flown ir such « manner that the pilot's eyes were held right on the ‘JASI _t1d@ slope all the way to touchdown, the main landing gear would touchdown quite near the point (300 to 365 feet beyond tie threshold) where Flight 505 did touchdowm. For example, calculations indicate that the main landing gear would contact the runway approximately 490 feet. short of the aiming point, or 400 fees. down this runway if the following representative conditions were assumed: distance from pilot aft to main danding gear of 70 feet; height of pilot's eyes above main landing gear tires of 14.8 feet; aircraft flown at 2.4° deck angle; anc at a deacent angle of 3.19. Another possible explanation for the short landing would be that the pilot performed a "duck-under" maneuver, This is a maneuver in which a pilot consciously positions his aircraft below the glide slope at a certain distance from: the runway threshold in order to permit an earlier flare to a landing, thereby giving himself more available runway on which to stop the aircraft. This maneuver, however, is inherently dangerous if not fully understood. The desceut below the original glide slope may require an appreciable increase in thrust to maintain the aircraft on a Nia OA Thoehge NAMIE RT pe alee ee et Reale a ec, oN AS, LN wt an ea po NE Gh tee meee Rm ase SOLS pa new and more shallow glide slopa to the desired touchdown point. If thrust is not increased, the aircraft will touchdown short of the desired touchdown point. Although the exact cause of the initial hard touchdown could not be determined, this landing did not cause catastrophic failure of the aixcraft, and it did not result in a subsequent uncontrollable maneuver. It is the opinion of the Board that, regardless of the physical or mental limitations imposed by the short runway and the surrounding hilly terrain, the pilot should have been able to recover from the bounce which followed the initial touchdown.

[ANALYSIS] Although the exact cause of the initial hard touchdown could not be determined, this landing did not cause catastrophic failure of the aixcraft, and it did not result in a subsequent uncontrollable maneuver. It is the opinion of the Board that, regardless of the physical or mental limitations imposed by the short runway and the surrounding hilly terrain, the pilot should have been able to recover from the bounce which followed the initial touchdown. Based upon its examina:ion of all evidence, the Board concludes that the high bounce vas more the result of pilot input to the controls than of eiastic rebound of the aircraft as a result of a very hard landing. This conclusion stems from the following facts: neither crewmembers nor passengers felt that the initial landing was excessively hard; the aircraft attained 4 great height (50 feet) on the first bounce; and the flighc engineer stated that the aircraft assumed an excessive pitch attitude as it began the ascent. Once this bounce occurred, the captain had two choices of action according to the Boeing Training Manual: (1) he could have completed the landing by performing the high bounce recovery technique; or (2) he could have executed a goraround to make a gecond approach. He attempted to Salvage the landing. However, possibly because of limitations imposed by the shoxt runway and the surrounding hilis, he modified the bounce recovery technique in «the following manner: (1) additional thrust was not applied to lessen the rate of descent during bounce recovery; and (2) the wing fiight spoilers were deployed while the aircraft was airborne. The timing of the spoiler actuation appears at first to he debatable. The flight engineer stated that he observed tha captain actuate the spoilers shortly after the aircraft crested the first: bounce. However, the initiation of the intermittent horn sound on the cockpit voice record did not occur wntil after the second touchdown. This horn ie the warning signal of an unsafe flight condition which is believed to have occurred because the spoiler ileve. was moved from the 0° detent while the flaps were extended. Since the timing of the horn and the flight angineer's statement seemed contradictory regarding the timing of spoiler actuation, the Board investigated further to resolve thig macter. The cause of this discrepancy appears to be the result of a 2-1/2 second delay between actuation of the spoiler lever and the initiation of the warning systenr. This delay was observed by a Board investigator in several similar model 727 aircraft. A delay of this magnitude would then place the actuation of the spoiler lever ac a time 1 second betore the circraft touched down the second time. fhe physical data also agree with that interpretacion. The impact of the tail skid at, or immediately after the second touchdown, indicates that the aircraft was operating at a high angle of attack at that tine.

[ANALYSIS] This delay was observed by a Board investigator in several similar model 727 aircraft. A delay of this magnitude would then place the actuation of the spoiler lever ac a time 1 second betore the circraft touched down the second time. fhe physical data also agree with that interpretacion. The impact of the tail skid at, or immediately after the second touchdown, indicates that the aircraft was operating at a high angle of attack at that tine. Thus, in order to set up the high descent rate reflected by this second touchdown in wpite of the relatively high angles of attack at which the aircraft must have been operating, it appears that the spoilers must have been used. Since the flight engineer stated that the spoilers were retracted shortly after their actuation, the Board can offer no reitson for the failure of the intermittent signal to stop before the horn switched to a continuous signal 4.5 seconds later, unless the spoiler lever was not placed back in the 0° detent. However, the continuous signal is a warning of an unsafe landing gear configuration which is believed to have actuated in tnia case because the final failure of the RMLG attsch structure broke the electrical connection to the landing year downlock switch. A question naturally arises concerning the reason ox reasons that control of the aircraft was lost. One <eagon considered was that the captain's seat locking mechaniam failei at the initial touchdown, and caused hin to strike hie head on some portion of the cabin interior. This b.low, it was reasoned, may have rendered the captain ¢éither unconscious or dazed, and this would account for his aubsequert loss of control of the aircraft and his loss of mmory. It was noted that the carrier hed experienced problems with the adjustment and locking mechanisus on the pilot seats in that model aircraft, and in this particular aircraft, in April 1970. Although the Board does not dispute the captain's statement that he suffered loss of memory, we cannot conclude that failure of his seat wag the cause of this. Among the reasons for our belief that failure of the captain's seat was not a factor in this accident are the following: (1) The results of the Board's tests indicate that Lateral wovernent of the upper torso waa required to permit the head to make contact with the cabin interior; however, the considerable lat:era) forces which would be required to cause this bodily movement were not likely generated in the initial touchdowm. Numerous passengers described the various impacts, and no one mentioned other than vertical forces in the initial touchdowm. High lateral accelerations would have kwen experienced, however, when the aircraft impacted the slope of Sara Fil. At that time, the aircraft was oriented approximately 90° tc its direction of motion. I , I Seat deficiensies of the type experienced by Trane Caribbean Airways and other users of that model seat wexe of the annoying, rather than the catastrophic, types they involved a limited amount of play in the weat rather than large ecale movement.

[ANALYSIS] In its analysis of the facts and circumstances of this accident, the Board assessed tie evidence bearing on the man-machine-environmental relationships. This approach led, in turn, to the formmlation of varicus hypotheses concerning the most probable causai areas of this accident. Tne first hypothesis considered ‘that @ destructively hard first touchdown occurred which was caused either by improper crew technigues or by external factors such as wind shear or turbulence. Another hypothesis considered that a mechanical failure which occurred sometime in the sequence of events was a direct cause of the accident. A final hypothesis is concerned with a breakdown in the interaction of the flightcrew with their aircraft subsequent to the initial touchdown. In the process of testing the3e hypotheses with observations made during the course of the investigation, the implication of the final hypothesis - the mnan/machine interface - became obvious. The factors which influenced the actions of the crew subsequent to the touchdown and the underlying factors prompted thes# events emerged as those of primary interest in determining the causal area of this accident. Before the third hypothesis is considered, the findings which disproved the first two hypotheses concerning possible cavaal areas will be discussed. Since the first causal axea presumes a hard initial touchdown as the direct cause of the accident, the nature of this landing must be reviewed. This first touchdown took place, according to witnesses, quite close to the end of the runway - approximately 300 feet beyond the threshold according to a controller in the tower, or 365 feet, if the tixe marks on the runway are accepted as those of this aircraft. This would place touchdown 435 to 500 feet prior te the VASI aiming point. The intensity of the touchdown was generally rated by surviving passengers as "hard", but not so hard as those following it. One witness described it as firm, ‘out not of an extreme nature. This witness was surprised at the height of the ascent that followed. In his statement, the captain described the landing as “very hard" and "very firm" - the flight engineer, as ". . . hard, definitely hard, but within safety bounds.% Tne accelevation trace of the fight data recorder confirms these Statements; the incremental accelerations recorded at the second and third touchdowns were beth approximately threa times that recorded at ths first touchiown. The physical evidence does not support the theory that the initial landing was catastrophic; the first evidence of scructural failure was located approximately 506 feet down the runway from the point of second touchdown. Finally, this theory is refuted by the lack of immediate concern shown by the crew. Only a few notuncommon remarks concerning the hard touchiown were made in the cockpit. It was not until slightly before the second touchdown when the voice record began to show a sense of {{mpending emergency in the tone cf voice and the comments made ky the crew.

[ANALYSIS] It was, perhaps, his preoccupation with these aspects of the approach maneuver which caused the captain to fiy the aircraft into a situation from which a short and hard landing was inevitable. Finding hingelf in a difficult situation immediately after the touchdow, seems to have confused the captain and affected hie further actions to salvage the landing. The Board feele that two factors may have combined to cause this .esponne s (1) the captain's lack of familiarity with the characteristics of the aircraft, in conjunction with the limitations imposed by the short runway at the Truman Airport, made him uncertain of the corrective action required; and (2) his power of reasoning was disrupted by natural behavioral changes which can occur in situations guch as that with which he was faced. The second factor concerns an individnal's natural respense to dangerous situations. Experiments conducted by Davis and reported in his study “Human Ferrers and Transport Accidents" 6/ explain the nature of this /cesponse. Davis noted that =man, like all animals, tndergoes certain behavioral changes) when danger appears imminent. These changes are intended to extract him rapidijy and impulsively from shat dangerous situation without having to go throug’ a slower reasoning process. In experiments in an artificial cockpit, Davis showed that this so-called emergency mechanism is detrimental in a situation which requires deliberate responses because it cancels the functioning of reasoning. These experiments showed that when a pecson reacts toward a situation in a way that experience (and training) have taught him to be effective, and that specific reaction deteriorates the situation instead, the emergency mechanism may set in within seconds. This creates confusion, which in turn, increases the sense of danger. A vicious circle is then formed which leads eitrer tc total inaction or to fruitless measures. In relating this theory to the circumstances at hand, it is intexesting to note that the captain's ~ttempts to salvage the landing by certain actions (such as an abrupt change in pitch attitude in the first place, and then by actuation of the spoilers during the bounce) only caused the ee ee OM Loe Ee ONT TE Tee ee Fe Te CE Eee er ee SE CE ee ere Oe A situation to deteriorate, contrary to his expectations. From that point on, che actions taken by the captain do not efem to be entirely rational. b. Post Crash Aspacts (1) Survivability Based upon the most common means of xneasurement, this was a survivable accident: The fuselage remained relatively intact; most of the occupants remaincd restiained; and the occupants had various means of immediate escape from the post-impact fire.

[ANALYSIS] At that time, the aircraft was oriented approximately 90° tc its direction of motion. I , I Seat deficiensies of the type experienced by Trane Caribbean Airways and other users of that model seat wexe of the annoying, rather than the catastrophic, types they involved a limited amount of play in the weat rather than large ecale movement. The nature of the design of the seat. locking mechanism is such that the Board concurs in tha manufacturer's statement that it cannot fail catastrophically. The locking mechanism is a self-contained unit which operates with the displecement of fluid thru & narrow orifice within the unit. Even with rapid loss of fluid, seat displacement would be gradual. SE Nee Se ee eae eee ee ee aN MMOS TR MURR TIER nr HEPE, To ven M GE ane ORE WEP Ep Meee aren Ss wtree vant Hn RT EA ON EOE SN Finally, remexvks made by the captain subsequent to the first towchdown and recorded on the CVR do not appear to the Board typical of those which would be expected from a person who was stunned by a blow to the head. About 3 eeconds after touchdawn, the captain made a remark commonly used in pilots’ parlance to express digeatisfaction with an event or situation. Rileo, hie Later commands for hie crew ta raisa the . flaps to takeoff position after he elected to ' go-around were lucid. The Board believes that the retrograde amnesia suffered by the captain was caused by the common defense mechanisin in wh:ch the system blocke certain traumatic expserienceas in orcer to allaviate psychic trauna. It is also possible that the blows on his head which occurred at final impact may have been the factor in initiating this amnesia. The events which followed the second touchdown reflect the increasing confusion which prevailed in the cockpit. With three men attempting ta control the aircraft, alternate periods of action and inaction resulted. thile these measures cannot ke considered causal factors, they did affect the severity of the accident. Ang, Maa BRE: Sanat, 1 I The second bounce wae a consequence of factors similar to those which caused the first bounce - touchdown at # high rate of descant with the aircraft operating at a high angle of attack. Again, thrust doea not appear to have been used to arrest the descent rate, and the final touchdiowa was also quite hard. The ferces generated at this tine completed the previously initiated failure of the gear attach structure. Afs geome time during the second bounce, the captain relinguished control of the aircraft to the first officer, who replied "I have ‘er." At that time, it is clear from the cockpit voice record that the crew wae aware of a developing eme::gency situation. Shortly afte the finas touchdown, the captain resumed control when he elected co execute a go-around.

[ANALYSIS] In this respect, certain aspects of the approach of Flight 505 seem noteworthy. « 27 - One such aspect is the somewhat reversed student ~ instructor relationship which developed in the cockpit @uring the approach. ‘This relationsnip was evident in the decision of the captain to experience the rasponse of the aircraft at slower speeds while on final approach. The factors influencing this decision will be discussed later in this report. Another noteworthy aspect of the approach was the protile fiown. Although the heading was flown rather precisely, the indicated airspeed was never really stabilized. The airspeed underwent a short-cycle variation of 5 knots above and 3 below reference sped during the latter portion of the approach. Throughout the first half of cne 3-minute final approach, the sink rate? was approximately 600 fpm. This increased to a relatively steady rate of descent of 680 fpm during the final 50 seconds. That rate of descent, in conjunction with the average airspeed during that period, corresponded to an average descent angle of 3.1°. The first officer noted that the sink rate was 700 fpm just 4.4 seconds before touchdown, and the flight data recorder: readout shows a constant vate Gascent down to its “sero" altitude. The Board does not consider the error in the VASI landing aid a significant factor of this accident. It should not have caused the crew to fly an approach angle steeper than the misaligned VASI setting of 2.75°. However, as has been noted, the aireraft actually flew a descent angle during the last portion of the approach which was significantly greater than that projected by the VASI aystem. The initial teuchdown must certainly be considered noteworthy. The Board did not arrive at any positive determination of the cause of this. Among the various causes considered possible were: a late fiare, an encounter with wind shear or turbulence, failure of the pilot to fiare af all, and performance of a "duck-under" maneuver kv the pilot. The possibility that the “ircraft was flared too late to arrest «he rate of descé... before touchdown seems j i ‘ i {{ ee plausible, especially in view of the stuep approach angle flown in this instance. Such a maneuver could also explain the high angle of attack the aircraft assumed upon becoming airborne again. The comments of the crew indicate that a wind shear might have caused the short touchdown, or that the aircraft encountered a downdraft just after it crossed the chreshold. However, no evidence that the flight encountered either phenomenon was found. The surface winds reported by the tower never exceeded 10 knots.

[CONCLUSIONS > FINDINGS] The captain’s overcontrol of the elevator during the landing and his failure to execute a go-around from a destabilized flare were causal to the accident. 11. The captain’s control inputs during the flare and bounce were not consistent with landing procedures and techniques outlined in the Federal Express MD-11 pilot training procedures, McDonnell Douglas flight crew operating manual, or with Federal Express’ MD-11 tailstrike awareness and high sink rate and bounce recovery training. 12. The captain had no previously documented skill deficiencies that contributed to this accident. Conclusions 70 Aircraft Accident Report 13. Air carrier pilots’ performance would be improved by additional guidance and training in landing techniques. 14. The flight crew’s calculation error in determining the runway length required for landing influenced the captain’s subsequent actions during final approach and landing by creating a sense of urgency to touch down early and initiate maximum braking immediately. 15. Some flight crewmembers may lack proficiency in the operation of airport performance laptop computers, or similar airplane performance computing devices, and confusion about calculated landing distances may result in potentially hazardous miscalculations of available runway distances after touchdown. 16. The inoperative left landing light did not impede the captain’s ability to land the airplane. 17. The MD-11's tendency to pitch up at ground spoiler deployment did not contribute to the accident. 18. The handling changes incorporated in the MD-11 flight control computer-908 software upgrade will provide valuable improvements in safety during MD-11 landings. 19. With the information that is currently available from the flight data recorder, it may be impossible to distinguish the control inputs of the MD-11 flight control computer-908 longitudinal stability augmentation system from the pilots’ control inputs. 20. The MD-11’s throttle resolver angle-driven spoiler knockdown feature did not contribute to this accident. 21. Additional basic research to identify undesirable landing phase combinations and to compare the overall qualitative and quantitative stability and control characteristics of widely used, large transport-category airplanes is needed to improve certification criteria and reduce the incidence of potentially catastrophic landing accidents. 22. The energy transmitted into the right main landing gear during the second touchdown was 3.2 times greater than the MD-11’s maximum certificated landing energy and was sufficient to fully compress (bottom) the right main landing gear strut and cause structural failure of the right wing rear spar. 23. The structural failure of the right wing rear spar resulted in the rupture of the right wing fuel tanks and fire. 24. The failure modes and effects for vertically fused and overdesigned landing gear designs may have been inadequately researched to identify whether, under overload conditions, one design might provide a safer break-up sequence for the airplane than the other design. Conclusions 71 Aircraft Accident Report 25.

[ANALYSIS] The pitch attitude was about 2° noseup and decreasing rapidly. The elevator was about 15° nose-down, although it was moving rapidly toward a nose-up position. Given the nose-down elevator position at that point in the bounce, there were probably no additional crew actions that could have been taken to prevent a hard impact with the runway. The airplane touched down for the second time as vertical acceleration was decreasing through 0.5 g. The second touchdown occurred at a roll angle of 9.5° right wing down, a roll rate of approximately 7° per second right wing down, and a pitch attitude of minus 0.7°. Peak vertical speed at the right MLG was approximately 13.5 fps. The right wing failed at impact (see section 2.5.1 for a discussion of this failure). The captain’s actions during the 5 seconds preceding the second touchdown established the conditions that led to the right wing failure. When the captain rapidly moved the elevators to near neutral instead of maintaining nose-up elevator and continuing the flare (2 seconds before first touchdown), he destabilized the flare and established a greater sink rate. The large nose-up elevator and thrust inputs that the captain made with only 1 second remaining before touchdown were his reaction to the sink rate and an attempt to prevent a hard landing. From that moment on, evidence indicates that all of the captain’s control inputs were too late and too large to achieve the desired effect. He made a large nose-down elevator input, consistent with an effort to keep the airplane on the runway and ensure an early touchdown of the nose gear with maximum available stopping distance. Although he began these nose-down inputs at about the time of the first touchdown, the airplane had bounced back into the air by the time he had pushed almost all the way forward on the control column. This large nose-down input, in turn, established a very high sink rate and low g load at the time of the second touchdown. The 98 The time interval between the first touchdown and the second was about 3 seconds. During those 3 seconds, the airplane was on the ground with the struts stroking and tires compressing for about 1 second and was airborne for about 2 seconds. 99 The Safety Board could not determine why the captain commanded right-wing-down aileron and left rudder deflection before the second touchdown. Analysis 54 Aircraft Accident Report captain’s final, large nose-up inputs were made too late to soften the impact. The airplane touched down with enough energy and at a sufficiently high roll angle to bottom the right MLG strut and break the right wing. All available data indicate that the airplane’s aerodynamic performance and flight control functionality were normal until after the second touchdown. Thus, the Safety Board concludes that the accident airplane performed normally in response to the captain's flight control inputs until after the second touchdown.

[ANALYSIS] A TWA L-1011 landed in New York at 14 fps, exhibiting vertical energy loads more than twice its certification requirements. Current landing phase structural design requirements only require consideration of 1.0 g vertical acceleration, small roll angles, and sink rates up to 12 fps. Manufacturers are also required to consider landing gear overloads in the up and aft directions but have the option of either fusing or overdesigning the gear for such loads. Several major landing accidents have now occurred as a result of pilots allowing their airplanes to land with more adverse combinations of lift, roll angle, and sink rate than those specified in the regulations. In each accident, a wing broke and a fuel fire erupted. Each of these accidents involved aircraft whose landing gear were not fused for upward (vertical) acting loads, which concerns the Safety Board. The Safety Board concludes that the failure modes and effects for vertically fused and overdesigned landing gear designs may have been inadequately researched to identify whether, under overload conditions, one design might provide a safer break-up sequence for the airplane than the other design. Therefore, the Analysis 67 Aircraft Accident Report Safety Board believes that the FAA should conduct a study to determine if landing gear vertical overload fusing offers a higher level of safety than when the gear is overdesigned. If fusing offers a higher level of safety, the FAA should revise 14 CFR Part 25 to require vertical overload fusing of landing gear. Further, peak vertical acceleration values recorded by the FDR at landing may not be sufficient for maintenance personnel to determine whether structural damage may have occurred during the landing. Data from the Newark accident indicate that initial vertical acceleration, pitch and roll rates, and attitudes should also be considered during FDR readout and evaluation of a potential hard landing event. The Safety Board notes that Boeing has revised its MD-11 maintenance manual to incorporate this guidance and that the company plans to revise the maintenance manuals of its other products based on the revised MD-11 maintenance manual example. However, the Board is concerned that this guidance will not be available to operators of non-Boeing products and that it is not binding. Thus, the Safety Board concludes that current manufacturer guidance for hard landing identification and operator maintenance readouts and analysis of FDR data following suspected hard landings may not be adequate to identify landings in which structural damage may have occurred. Therefore, the Safety Board believes that the FAA should require manufacturers of 14 CFR Part 23 and Part 25 airplanes and Part 121 operators to revise their hard landing inspection and reporting criteria to account for all factors that can contribute to structural damage.

[ANALYSIS] Aviation Safety and Pilot Control: Understanding and Preventing Unfavorable Pilot-Vehicle Interactions, p. 3. Analysis 55 Aircraft Accident Report 2.2.2 Flight Crew Factors During the Approach and Landing During the approach briefing, the first officer and the captain discussed the stopping distance available on runway 22R for the airplane’s weight and landing configuration. During that discussion, they expressed concerns about the approximate landing distance and the length of the runway, which they had derived from the APLC (see section 2.2.2.7 for a discussion of the flight crew’s misinterpretation of the data presented in the APLC). Additionally, during the approach, the flight crew indicated that they were aware of the inoperative No. 1 engine thrust reverser, which would have resulted in a slight reduction in deceleration capability after landing.101 The flight crew was also aware of three recent events recorded in the airplane’s maintenance log in which the airplane’s autobrakes had failed to arm at takeoff or failed to work at landing. Although maintenance personnel had checked the system after each reported failure and determined it was functioning properly, the captain told Safety Board investigators that he discussed the reliability of the autobrake system with the first officer before takeoff from ANC. The captain told investigators that the autobrakes remained armed during the departure from ANC. However, he kept the autobrake problem in mind when planning for the landing at EWR, adding that he planned to land the airplane at the start of the runway and wanted to ensure that the airplane would not float during the landing flare. Thus, on the basis of the flight crew’s comments during the approach about the relatively short runway length, the inoperative thrust reverser, the questionable reliability of the autobrake system, and the perceived need to land at the beginning of the runway, the Safety Board concludes that the captain was concerned about the airplane’s touchdown location on runway 22R and intended to take measures during the landing to achieve an early touchdown and minimize the length of the rollout on the runway after touchdown. 2.2.2.1 Nose-Down Elevator Input at 0132:16 (2 seconds before first touchdown) The Safety Board examined the captain’s 12° nose-down elevator input at 17 feet radio altitude to determine if it was consistent with FedEx guidance for landing the MD-11. The Board’s review of FedEx’s MD-11 landing guidance found only one technique that promotes the use of nose-down elevator between the initiation of flare and touchdown. Specifically, the FedEx MD-11 “advanced technique” for landing recommends that “elevator back pressure…be relaxed” about 10 feet before touchdown (to achieve a 1° decrease in pitch attitude). However, the captain’s nose-down elevator input, which moved the elevator from 12° nose-up to about the neutral position, was very rapid and much greater than is required for the maneuver.

[CONCLUSIONS] The structural failure of the right wing rear spar resulted in the rupture of the right wing fuel tanks and fire. 24. The failure modes and effects for vertically fused and overdesigned landing gear designs may have been inadequately researched to identify whether, under overload conditions, one design might provide a safer break-up sequence for the airplane than the other design. Conclusions 71 Aircraft Accident Report 25. Current manufacturer guidance for hard landing identification and operator maintenance readouts and analysis of flight data recorder data following suspected hard landings may not be adequate to identify landings in which structural damage may have occurred. 26. Risks to firefighters and the surrounding community were minimized substantially because the incident commander assumed that hazardous materials were on board and acted accordingly. 27. The Newark accident demonstrates that air carriers transporting hazardous materials continue to need a means to quickly retrieve and provide consolidated, specific information to emergency responders about the identity of all hazardous materials on an airplane. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the captain’s overcontrol of the airplane during the landing and his failure to execute a go-around from a destabilized flare. Contributing to the accident was the captain’s concern with touching down early to ensure adequate stopping distance. 72 Aircraft Accident Report 4. Recommendations 4.1 New Recommendations As a result of the investigation of this accident, the National Transportation Safety Board makes the following recommendations: To the Federal Aviation Administration: Convene a joint government-industry task force composed, at a minimum, of representatives of manufacturers, operators, pilot labor organizations, and the Federal Aviation Administration to develop, within 1 year, a pilot training tool to do the following: Include information about factors that can contribute to structural failures involving the landing gear, wings, and fuselage, such as design sink rate limits; roll angle limits; control inputs’ roll rate; pitch rate; single-gear landings; the effect of decreased lift; and structural loading consequences of bottoming landing gear struts and tires; (A-00-92) Provide a syllabus for simulator training on the execution of stabilized approaches to the landing flare, the identification of unstabilized landing flares, and recovery from these situations, including proper high sink rate recovery techniques during flare to landing, techniques for avoiding and recovering from overcontrol in pitch before touchdown, and techniques for avoiding overcontrol and premature derotation during a bounced landing; (A-00-93) and Promote an orientation toward a proactive go-around. (A-00-94) Require principal operations inspectors assigned to Part 121 carriers that use auxiliary performance computers to review and ensure the adequacy of training and procedures regarding the use of this equipment and the interpretation of the data generated, including landing distance data. (A-00-95) Require the installation, within 1 year, of the MD-11 flight control computer-908 software upgrade on all MD-11 airplanes. (A-00-96)

[ANALYSIS] Therefore, the Safety Board concludes that the captain’s full nose-down elevator control input at the time of the first touchdown was consistent with his continued concerns to avoid a long landing and his desire to avoid a tailstrike. 2.2.2.4 Summary of the Captain’s Elevator Control Inputs Considering the captain’s three significant elevator control inputs in sequence, it is apparent that after the first destabilization of the landing flare (from the captain’s nosedown input at 17 feet agl), each of the succeeding nose-up/nose-down elevator inputs resulted from the captain’s attempt to correct for the immediately preceding control input. His perception of a short runway and the need to constrain the pitch attitude within a very limited range (to avoid a tailstrike) would have motivated the captain to rapidly return the airplane to a stable attitude. He attempted to accomplish this goal with the quick application of large elevator inputs; however, this succession of elevator inputs and pitch oscillations rendered the landing attempt increasingly unstable. Throughout the sequence of increasingly extreme nose-down and nose-up elevator inputs, which were consistent with a “classic” PIO (as described in section 2.2.1), the captain continued to attempt to salvage the landing; however, a go-around executed by the captain at any time through the touchdown and bounce would have prevented the accident. 102 MD-11 pilots are taught and the MD-11 FCOM advises that ground spoiler deployment at touchdown creates a nose-up pitching moment that must be counteracted with pilot-induced nose-down elevator inputs. This technique is referred to in the MD-11 FCOM and operator training as “flying the nose to the runway.” Analysis 58 Aircraft Accident Report Therefore, the Safety Board concludes that the captain’s overcontrol of the elevator during the landing and his failure to execute a go-around from a destabilized flare were causal to the accident. Further, the Safety Board’s examination of the training that FedEx provided its pilots in landing the MD-11 showed that its training was consistent with and, in some respects, exceeded that provided by many other major airlines. On the basis of comparing the captain’s control inputs with FedEx’s procedures and training for landing the MD-11, the Safety Board concludes that the captain’s control inputs during the flare and bounce were not consistent with landing procedures and techniques outlined in the FedEx MD-11 pilot training procedures, McDonnell Douglas FCOM, or with FedEx’s MD-11 tailstrike awareness and high sink rate and bounce recovery training. 2.2.2.5 The Captain’s Training History The Safety Board attempted to determine if a factor in the captain’s training history could explain his actions in attempting to control the airplane during the landing and thereafter.

[ANALYSIS] Thus, the Safety Board concludes that the captain’s execution of the beginning of the flare maneuver was normal and not a factor in the accident. As pitch attitude peaked about 2 seconds before the first touchdown, the elevator started deflecting from about 12° nose up to near 0°, and the airplane’s pitch attitude began decreasing slightly in response to the nose-down elevator input. Further, about 1 second before ground contact, elevator deflection reversed to a nose-up elevator deflection of 26° (from about neutral elevator to about 70 percent of maximum nose-up elevator), and TRAs increased from about 40° to 70° (from near flight idle to near takeoff thrust). A small right-wing-down aileron input (4° to 5°) followed. The nose-up, throttleup, and right-wing-down control inputs were initiated as the airplane was descending through 7 feet radio altitude. Pitch attitude and vertical acceleration had just begun to respond when the airplane contacted the ground in the first of two touchdowns. Vertical speed at the first touchdown was about 7.6 fps,97 and vertical acceleration peaked at 1.67 g. The nose-up elevator and throttle inputs also peaked about the time of the first touchdown. 96 With the MD-11 autothrottle system engaged and flaps extended to greater than 31.5°, the throttles are automatically driven to the idle stop when the radio altitude decreases through 50 feet. 97 This value is the vertical speed at the right MLG and includes 6.6 fps vertical speed at the c.g. plus 1.0 fps vertical speed at the right MLG because of nose-up pitch rate and right-wing-down roll rate. Analysis 53 Aircraft Accident Report Within 1/2 second after the first touchdown, the captain initiated a rapid nosedown elevator input. The total elevator travel was about 40° (changing from about 70 percent of maximum nose-up elevator to about 67 percent of maximum nose-down elevator in less than 1 second). Despite the initiation of the large and rapid nose-down elevator input, the airplane began to lift off the runway as a result of landing gear strut and tire compression loads and the still-increasing pitch attitude, thrust, and airspeed. In addition, wing lift was not degraded upon touchdown because the spoilers did not deploy. After the initial touchdown, the airplane was airborne for about 2 seconds.98 During the first second, while airborne, the elevator remained about 67 percent nose down. In the next second, a large and rapid nose-up elevator input occurred (from 67 percent nose-down to 60 percent nose up), accompanied by nose left rudder and right-wing-down aileron inputs.99 About 3/4 second before the second touchdown, as the airplane was peaking at a height of 5 feet agl, lift had decreased to about 0.6 g. The pitch attitude was about 2° noseup and decreasing rapidly. The elevator was about 15° nose-down, although it was moving rapidly toward a nose-up position.

[ANALYSIS] Include information about factors that can contribute to structural failures involving the landing gear, wings, and fuselage, such as design sink rate limits; roll angle limits; control inputs’ roll rate; pitch rate; single-gear landings; the effect of decreased lift; and structural loading consequences of bottoming landing gear struts and tires; b. Provide a syllabus for simulator training on the execution of stabilized approaches to the landing flare, the identification of unstabilized landing flares, and recovery from these situations, including proper high sink rate recovery techniques during flare to landing, techniques for avoiding and recovering from overcontrol in pitch before touchdown, and techniques for avoiding overcontrol and premature derotation during a bounced landing; and c. Promote an orientation toward a proactive go-around. 2.2.2.7 Landing Distance Calculation Errors During its investigation, the Safety Board determined that the flight crew misinterpreted the APLC stopping distance data for MED autobrakes by incorrectly comparing APLC runway data with the landing distance provided on the approach plate for runway 22R. Although there was sufficient stopping distance for a MED autobrake setting, the misinterpretation of the APLC data,105 among other factors, led the captain to believe that stopping distance would be an issue in the landing. Thus, the Safety Board concludes that the flight crew’s calculation error in determining the runway length required for landing influenced the captain’s subsequent actions during final approach and landing by creating a sense of urgency to touch down early and initiate MAX braking immediately. The Safety Board is concerned that two pilots with significant APLC experience at FedEx failed to properly interpret the calculated landing distances and that other experienced flight crews may also be deficient in their operational knowledge of how APLC systems function. The Board notes that following the accident, FedEx expanded the APLC pilot training presentation for all initial and upgrade training and also added it to recurrent flight crew training programs. The Board has learned that several operators have either adopted systems similar to FedEx’s APLC system or are considering doing so and that other electronic performance calculators are in use at other operators. Thus, the Safety Board concludes that some flight crewmembers may lack proficiency in the operation of APLCs, or similar airplane performance computing devices, and that confusion about calculated landing distances may result in potentially hazardous miscalculations of available runway distances after touchdown. Therefore, the Safety Board believes that the 105 Instead of the miscalculated 780-foot margin result that influenced his decision to set MAX autobrakes, there was actually a 1,680-foot margin. Analysis 61 Aircraft Accident Report FAA should require POIs assigned to Part 121 carriers that use auxiliary performance computers to review and ensure the adequacy of training and procedures regarding the use of this equipment and the interpretation of the data generated, including landing distance data. 2.2.2.8 Left Landing Light The Safety Board considered whether the absence of the left landing light affected the captain’s ability to land the airplane.

[ANALYSIS] The response and actions by ARFF and area firefighting personnel were timely and adequate, despite a lack of timely information about the nature and quantity of hazardous materials on board. Clear night visual meteorological conditions with light winds prevailed at the time of the accident; weather was not a factor in the accident. This analysis examines airplane and flight crew performance and design and certification requirements for the performance of transport-category airplanes during the landing phase. The analysis concludes with an examination of problems encountered during the emergency response relating to the timely dissemination of hazardous materials cargo information. 52 Aircraft Accident Report 2.2 Accident Scenario 2.2.1 Airplane Performance During the Approach and Landing During the approach, the airplane was configured for landing, with flaps set at 50°. The captain disconnected the autopilot as the airplane descended through 1,200 feet, and the autothrottles remained engaged. According to flight crew statements and FDR data, the airplane maintained the approach speed of about 158 knots (consistent with the target approach speed specified by FedEx, Vref+5 knots or 157 knots), at a stable 800 fpm descent rate, and on the ILS localizer and glideslope for runway 22R until the landing flare. The average pitch attitude of 3° ANU was consistent with MD-11 flight manual data for descending on the ILS glideslope’s 3° flightpath angle, given the airplane’s weight, c.g., and flaps-50 configuration. The captain and the first officer also stated that the approach was routine until just before touchdown. Thus, on the basis of flight crew statements and airplane performance data, the Safety Board concludes that the airplane’s approach before the landing flare was stabilized. FDR data indicated that control inputs consistent with the start of flare occurred at about 37 feet radio altitude. Engine thrust was also decreasing about this time.96 About 1.5 seconds after the start of the flare and 2 seconds before the first of two touchdowns, pitch attitude peaked at 5° nose up. The radio altitude was 17 feet. This portion of the flare maneuver was consistent with FedEx MD-11 flight manual guidance, which called for a “smooth 2.5 degree flare” to be initiated between 30 and 40 feet radio altitude. Thus, the Safety Board concludes that the captain’s execution of the beginning of the flare maneuver was normal and not a factor in the accident. As pitch attitude peaked about 2 seconds before the first touchdown, the elevator started deflecting from about 12° nose up to near 0°, and the airplane’s pitch attitude began decreasing slightly in response to the nose-down elevator input.

[ANALYSIS] Further, DC-10 and MD-11 training and procedures require pilots to manually deploy ground spoilers if they do not automatically deploy. Therefore, the Safety Analysis 64 Aircraft Accident Report Board concludes that the MD-11’s TRA-driven spoiler knockdown feature did not contribute to this accident. Nevertheless, the Safety Board notes that it is possible to modify the existing DC-10 and MD-11 spoiler deployment system to allow greater throttle movement before the spoiler knockdown feature is activated. Delaying the knockdown feature would allow pilots to make larger thrust increases just before landing without preventing ground spoiler deployment at touchdown, which may help prevent or minimize some bounces. In the event of a go-around, the higher knockdown angle would slightly delay the retraction of ground spoilers; therefore, a study to determine an optimum angle for activation of the knockdown feature would be necessary. Therefore, the Safety Board believes that the FAA should review and, if appropriate, revise the DC-10 and MD-11 TRA-driven ground spoiler knockdown feature to ensure that it does not prevent ground spoiler deployment at moderate TRAs that could be associated with sink rate and airspeed corrections during the landing phase. Further, the Safety Board believes that the FAA should require DC-10 and MD-11 operators to provide their pilots with information and training regarding the ground spoiler knockdown feature and its effects on landing characteristics and performance. 2.4 Transport-Category Airplane Stability and Control During the Landing Phase The records of previous MD-11 accidents and incidents (reviewed in sections 1.18.4 through 1.18.7), including the accident airplane’s two hard landing events that preceded the Newark accident, have drawn attention specifically to the landing characteristics of the MD-11. However, other transport airplane types, including the Boeing DC-10 and 757/767 (as cited by the Safety Board in its June 16, 1994, safety recommendation letter), also have been involved in landing accidents that were or could have been catastrophic. Although improved pilot training in landing techniques and installation of the FCC-908 software upgrade can help prevent MD-11 landing incidents and accidents (see sections 2.2.2.6 and 2.3.2), the accident history involving the MD-11 and other transport airplane types prompted the Board to consider and review existing certification criteria for airplane handling qualities during landing operations. The review indicated that, besides basic stability criteria, few objective standards exist for the assessment and acceptance of these handling qualities, including the interactions of airplane and pilot responses and the effects of adverse environmental conditions. Based on the accident and incident record, the Safety Board is concerned that certain complex system interactions, pilot input characteristics, and other factors, such as c.g. position and atmospheric conditions, may occasionally combine during the landing phase in undesirable ways that were not identified during the original certification of transport airplanes.

[ANALYSIS] Specifically, the FedEx MD-11 “advanced technique” for landing recommends that “elevator back pressure…be relaxed” about 10 feet before touchdown (to achieve a 1° decrease in pitch attitude). However, the captain’s nose-down elevator input, which moved the elevator from 12° nose-up to about the neutral position, was very rapid and much greater than is required for the maneuver. Further, the captain began his nose-down input about 1 second before the airplane reached 10 feet radio altitude, the aural annunciation of which should have served as the cue for such a pitch reduction if it 101 Although the flight crew may have been concerned about the reduction in deceleration capability, the inoperative thrust reverser did not increase the runway length requirement for the accident landing above that shown in the APLC because the deceleration effects of the thrust reversers are not used in calculating the distances required for landing. Analysis 56 Aircraft Accident Report had been related to the FedEx “advanced” landing technique. Thus, the Safety Board concludes that the captain’s nose-down elevator input beginning at 17 feet radio altitude was not consistent with FedEx guidance for landing the MD-11. Further, the Safety Board concludes that the captain’s nose-down elevator input at 17 feet radio altitude (2 seconds before the first touchdown) was consistent with an attempt to control the point of touchdown given his concerns about the runway length. 2.2.2.2 Nose-up Elevator Input at 0132:17 (1 second before first touchdown) The captain and the first officer told Safety Board investigators that they felt the airplane’s sink rate increase shortly before the airplane touched down. They stated that these were “seat of the pants” feelings and were not based on observed indications on cockpit instruments. FDR data indicated that after the captain made the nose-down elevator input at 17 feet radio altitude, a small increase in sink rate and decrease in vertical acceleration occurred. The decreased vertical acceleration and increased nose-down pitch rate could have led to sensations of sink consistent with the pilots’ descriptions. With just more than 1 second remaining before touchdown, the captain had the following options: accept the sink rate and subsequent hard landing, attempt to salvage the landing with last-second thrust and pitch adjustments, or execute a go-around. FDR data and postaccident interviews show that the captain chose to try to salvage the landing with last-second thrust and pitch adjustments. Thus, the Safety Board concludes that the captain made a nearly full nose-up elevator input and a large throttle increase to compensate for the increased sink rate caused by his previous nose-down input. The FedEx MD-11 flight manual recommends that a “constant pitch attitude be maintained from 10 feet radio altitude until touchdown.” However, this guidance presupposes a stabilized approach and flare leading up to 10 feet radio altitude. In contrast, because the captain had destabilized the flare 1 second earlier, he perceived a need to arrest the resulting sink rate with additional thrust and nose-up pitch.

[ANALYSIS] Energy not absorbed by the landing gear was then transmitted to the right wing rear spar through the right MLG attach points. A corresponding down load was introduced from the left wing and fuselage, which produced additional torsional loads on the right wing. These torsional loads then produced a shear overload condition in the right wing rear spar according to MDI/Boeing simulations. Boeing stated that the MDI simulations indicate that the failure most probably “initiated at the rear spar/bulkhead (trunnion) rib interface and progressed through the primary wing box structure. As a result of this failure, the right MLG trunnion moved substantially upward and aft with respect to the trap [trapezoidal] panel fitting.” Thus, the Safety Board concludes that the energy transmitted into the right MLG during Analysis 66 Aircraft Accident Report the second touchdown was 3.2 times greater than the MD-11’s maximum certificated landing energy and was sufficient to fully compress (bottom) the right MLG strut and cause structural failure of the right wing rear spar. Runway sooting consistent with a fuel fire near the right MLG was also found at the area of the second touchdown. Thus, on the basis of runway evidence, analysis of performance data, and the MDI/Boeing structural simulations, the Safety Board concludes that the structural failure of the right wing rear spar resulted in the rupture of the right wing fuel tanks and fire. Although the Hong Kong Civil Aviation Department’s investigation of the August 22, 1999, China Airlines MD-11 accident at Hong Kong International Airport is ongoing, examination of the pertinent fracture surfaces suggests that the wing spar failure mode in this accident was very similar to that of FedEx flight 14. The Safety Board’s preliminary calculations of the China Airlines MD-11’s descent rate at impact, 18 to 20 fps, imply that like the FedEx accident at Newark, the wing spar failed in overload well in excess of certification requirements. 2.5.2 Landing Gear Certification The MD-11 MLG was designed to break from the wing (fuse) in a drag overload condition but not in a vertical overload condition. Boeing has stated that this design was implemented because data indicated that the most likely landing gear overload condition would occur as a result of striking an obstruction. This “sacrificial shedding” of MLG assemblies in the aft direction was intended to prevent catastrophic loads being transmitted to the wing box and causing rupture. During its investigation of the FedEx Newark accident, the Safety Board reviewed the circumstances of several accidents involving other wide-bodied airplane types that greatly exceeded aircraft structural limits. A Martinair DC-10 touched down at Faro, Portugal, with a sink rate of 17 fps, at vertical energy loads 2.6 times greater than energy certification requirements for a single MLG. A TWA L-1011 landed in New York at 14 fps, exhibiting vertical energy loads more than twice its certification requirements. Current landing phase structural design requirements only require consideration of 1.0 g vertical acceleration, small roll angles, and sink rates up to 12 fps.

[ANALYSIS] The Safety Board’s review of accidents involving pilots’ control handling in the landing phase of flight, including this accident, indicates that a similar training tool development effort should be made for landings. This tool should devote specific attention to proper high sink rate recovery techniques during the landing flare, risks associated with PIOs during the landing, and the hazards associated with overcontrol and premature derotation during a bounced landing. In 1995, responding to a safety recommendation issued by the Safety Board as a result of its investigation of three Boeing 767 landing accidents as well as incidents involving DC-10s and MD-11s, the FAA issued FSAT 95-06. This document required FAA POIs to ensure that pilot training programs for the Boeing 757/767, DC-10, and MD-11 include a discussion about derotation accidents. Unfortunately, FSAT 95-06 expired in 1996. Further, in its submission to the Safety Board on the Newark accident, Boeing advocated expanding traditional approach go-around guidance to instruct that missed approaches be made if the airplane is not stabilized by 500 feet or if approaches involve “large pitch deviations.” The Board concurs with this suggestion and notes that air carrier pilots’ adoption and use of a proactive go-around philosophy would be a desirable goal for a training tool development effort on this issue. Following this accident, FedEx added instructional material and guidance on landing gear and wing structural certification to its tailstrike awareness training program. This guidance detailed the effects of vertical acceleration on the MLG and wings and explained the effects of roll and pitch rate on total sink rate. The FedEx training information describes in detail the aerodynamic effects of large nose-down elevator inputs that result in reduced-g touchdowns, which increase the loads that must be absorbed by the MLG. The Safety Board notes that one of the new FedEx training modules closely describes the acceleration, pitch, and roll factors found in the Newark accident scenario. However, based on discussions with pilots who have flown with several air carriers, the Board is concerned that this information may be lacking in other operators’ training programs and that this lack of landing guidance could contribute to similar landing accidents. Thus, based on its review of air carrier landing accidents, the Safety Board concludes that air carrier pilots’ performance would be improved by additional guidance and training in landing techniques. 104 A review of the Safety Board’s database of U.S. accidents revealed no fatal overrun events since 1990. Analysis 60 Aircraft Accident Report Therefore, the Safety Board believes that the FAA should convene a joint government-industry task force composed, at a minimum, of representatives of manufacturers, operators, pilot labor organizations, and the FAA to develop, within 1 year, a pilot training tool to do the following: a. Include information about factors that can contribute to structural failures involving the landing gear, wings, and fuselage, such as design sink rate limits; roll angle limits; control inputs’ roll rate; pitch rate; single-gear landings; the effect of decreased lift; and structural loading consequences of bottoming landing gear struts and tires; b.

[ANALYSIS] Based on the accident and incident record, the Safety Board is concerned that certain complex system interactions, pilot input characteristics, and other factors, such as c.g. position and atmospheric conditions, may occasionally combine during the landing phase in undesirable ways that were not identified during the original certification of transport airplanes. Thus, the Safety Board concludes that additional basic research to identify undesirable landing phase combinations and to compare the overall qualitative and quantitative stability and control characteristics of widely used, large transport- Analysis 65 Aircraft Accident Report category airplanes is needed to improve certification criteria and reduce the incidence of potentially catastrophic landing accidents. Therefore, the Safety Board believes that the FAA should sponsor a National Aeronautics and Space Administration (NASA) study of the stability and control characteristics of widely used, large transport-category airplanes to a. Identify undesirable characteristics that may develop during the landing phase in the presence of adverse combinations of pilot control inputs, airplane c.g. position, atmospheric conditions, and other factors; and b. Compare overall qualitative and quantitative stability and control characteristics on an objective basis. The study should include analyses of DC-10 and MD-11 landing accidents and any other landing incidents and accidents deemed pertinent by NASA. Further, the Safety Board believes that, based on the results of the study, the FAA should implement improved certification criteria for transport-category airplane designs that will reduce the incidence of landing accidents. 2.5 Structures 2.5.1 Right-Wing Structural Design and Failure Title 14 CFR Part 25 requires that an airplane’s landing gear and associated structure be able to withstand a 12 fps vertical speed when landing at maximum landing weight on one gear at zero roll angle and 1.0 g lift. This equates to a maximum energy capacity for a single MD-11 MLG, as required for certification, of 494,500 ft-lbs. Boeing estimates that the MD-11 landing gear strut will bottom and cause the wing rear spar to fail if approximately more than 1,500,000 ft-lbs of energy is transmitted into a single MLG. At 13.5 fps vertical speed, 0.5 g vertical acceleration, and 8° roll angle, the accident airplane’s right MLG experienced an energy input of 1,574,000 ft-lbs during the second touchdown, which was 3.2 times the maximum certification energy and slightly greater than the MD-11’s estimated ultimate capability. The MDI/Boeing structural simulations of the accident sequence indicate that the right MLG strut and outboard tires bottomed at the second touchdown. Energy not absorbed by the landing gear was then transmitted to the right wing rear spar through the right MLG attach points. A corresponding down load was introduced from the left wing and fuselage, which produced additional torsional loads on the right wing. These torsional loads then produced a shear overload condition in the right wing rear spar according to MDI/Boeing simulations.

[ANALYSIS] The Board notes that the captain received an unsatisfactory evaluation on an upgrade proficiency checkride on October 29, 1996. However, the Board obtained no other evidence that could reflect negatively on the captain’s skills. Other than the October 1996 checkride, there was no history of unsatisfactory performance or of disciplinary action in his career at FedEx. There was also no record of accident, incident, or enforcement action in his FAA records. In addition, in the 10 months after the failed checkride, the captain satisfactorily completed a proficiency check and two line checks (the last line check was 20 days before the accident). Thus, the Safety Board concludes that the captain had no previously documented skill deficiencies that contributed to this accident. 2.2.2.6 Enhanced Pilot Training The captain’s failure to properly respond to a destabilized flare and his excessive overcontrol of the airplane, as well as the accumulated evidence from previous air transport landing accidents (see sections 1.18.4 through 1.18.7), indicate that action may be warranted to improve the quality of air carrier training and guidance to pilots in performing safe landings. The circumstances of this and other accidents suggest that, although accidents before or shortly after touchdown are rare, the risk of a future catastrophic accident could be reduced if air carrier pilot training programs devote additional attention to safety issues related to landings. It is particularly important to instill in pilots the orientation to perform a go-around in the event of an unstabilized approach or destabilized landing flare. Shortly after the Safety Board conducted a special investigation103 of rejected takeoff accidents in 1990, a joint government-industry task force was formed to study the issue and develop a flight crew training aid. This training aid has led to a reduction in the 103 National Transportation Safety Board. 1990. Runway Overruns Following High Speed Rejected Takeoffs. Special Investigation Report NTSB/SIR-90-02. Washington, D.C. Analysis 59 Aircraft Accident Report incidence of rejected takeoff accidents and incidents.104 The Board notes that other government-industry efforts have produced valuable training tools to avoid and recover from inadvertent encounters with wake vortices, windshear, controlled flight into terrain events, and aircraft upsets. The Safety Board’s review of accidents involving pilots’ control handling in the landing phase of flight, including this accident, indicates that a similar training tool development effort should be made for landings. This tool should devote specific attention to proper high sink rate recovery techniques during the landing flare, risks associated with PIOs during the landing, and the hazards associated with overcontrol and premature derotation during a bounced landing.

[ANALYSIS] The FedEx MD-11 flight manual recommends that a “constant pitch attitude be maintained from 10 feet radio altitude until touchdown.” However, this guidance presupposes a stabilized approach and flare leading up to 10 feet radio altitude. In contrast, because the captain had destabilized the flare 1 second earlier, he perceived a need to arrest the resulting sink rate with additional thrust and nose-up pitch. FedEx’s high sink rate and bounce recovery training recommends establishing a 7.5° pitch attitude and “arresting the sink rate with thrust” as a prelude to either landing with a high sink rate, re-landing the airplane after a bounce, or executing a low-level goaround. However, FedEx’s MD-11 tailstrike awareness training also cautioned that “quickly adding up elevator” near the ground should be avoided because it can result in increased nose-up pitch rate at touchdown, increased downward vertical speed at the MLG, a hard landing, and tailstrike. To gain a better understanding of this training and its relevance to the captain’s actions, Safety Board investigators participated in FedEx classroom and simulator training for high sink rate and bounce recovery, as well as for tailstrike avoidance. This experience demonstrated to investigators that the timing and large magnitude of the captain’s nose-up elevator input just before the first touchdown were inconsistent with FedEx’s MD-11 high sink rate recovery and tailstrike awareness training. Analysis 57 Aircraft Accident Report 2.2.2.3 Nose-Down Elevator Input Shortly After the First Touchdown The captain’s large, nose-down elevator input began within 1/2 second of the first touchdown. Based on the sequence and timing of the events, this nose-down elevator input was the captain’s response to the airplane’s rapid nose-up pitching motion, which began in the second before touchdown as a result of the captain’s immediately preceding large noseup elevator input, and/or his attempt to rapidly land the nosewheel and begin braking immediately after touchdown. After the airplane touched down hard and bounced, the captain continued his nose-down input while the airplane continued to pitch up. A large nose-up pitch rate and high pitch attitude at touchdown would have introduced several factors that may have contributed to the captain’s subsequent large nose-down elevator input. First, MD-11 pilots are taught in training that nose-up pitch rate and high pitch attitude at touchdown are factors that lead to tailstrike. This consideration may have caused the captain to believe he should apply additional nose-down elevator to the amount that he normally applies after touchdown to counter the MD-11’s characteristic nose-up pitching moment following ground spoiler deployment.102 Second, as demonstrated by his statements on the CVR and during postaccident interviews, the captain would have continued to be concerned about the available runway length; the rapidly increasing pitch attitude just before and during the first touchdown would have increased the probability of a floating flare, which, in turn, would have decreased the amount of runway available to bring the airplane to a stop.

[ANALYSIS] Analysis 62 Aircraft Accident Report 2.3 MD-11 Handling Characteristics and Flight Control System Design 2.3.1 MD-11 Nose-Up Pitching Moment Because of Ground Spoiler Deployment The MD-11’s known tendency to pitch up after ground spoiler deployment and the captain’s reference to it during interviews prompted the Safety Board to evaluate the role of the pitch-up tendency in the accident sequence. The captain told Board investigators that he was expecting the nose-up pitching moment associated with initial spoiler deployment at MLG spin-up. He stated that he remembered compensating with forward control column input and that he thought the spoilers had deployed at touchdown. Although a portion of the captain’s nose-down elevator input at the time of the first touchdown may have been in response to the pitch-up tendency, the input greatly exceeded that required to control this tendency. Therefore, the Safety Board concludes that the MD-11's tendency to pitch up at ground spoiler deployment did not contribute to the accident. Nevertheless, a reduction or elimination of the pitch-up tendency would simplify MD-11 landing techniques and may help prevent future MD-11 landing incidents and accidents. 2.3.2 MD-11 Pitch Handling Characteristics and the FCC-908 Software Upgrade The FCC-908 software package developed by Boeing will alter the handling of the airplane during landings by decreasing the pitch sensitivity through action of the PRD. The decrease in pitch sensitivity combined with additional handling improvements included in the FCC-908 upgrade should render the airplane less susceptible to overcontrol in pitch similar to that involved in this accident. Boeing’s stated goal in implementing FCC-908 is to match the handling characteristics of the MD-11 to those of the existing DC-10 and the DC-10’s newly developed two-pilot adaptation, the MD-10, thereby facilitating FAA approval of a common type rating for the MD-10 and MD-11. The DC-10 and MD-10 do not have the pitch sensitivity that, until implementation of the FCC-908 software upgrade, has been characteristic of the MD-11. Further, the MD-11 FCC-908 software upgrade may help prevent tailstrikes by providing PAP and eliminating the MD-11’s nose-up pitching tendency at touchdown through the positive nose-lowering feature of FCC-908. The Safety Board notes that changing the FCC software to eliminate the nose-up pitching tendency may be an acceptable alternate approach to changing MD-11 spoiler logic as recommended in Safety Recommendation A-93-59. The Safety Board concludes that the handling changes incorporated in the MD-11 FCC-908 software upgrade will provide valuable improvements in safety during MD-11 landings. Therefore, the Safety Board believes that the FAA should require the Analysis 63 Aircraft Accident Report installation, within 1 year, of the MD-11 FCC-908 software upgrade on all MD-11 airplanes. 2.3.3 Digital Flight Data Recorder Update Required by FCC-908 The Safety Board notes that for an MD-11 equipped with the FCC-908 software package, the LSAS will apply elevator control inputs simultaneous with those of the pilots.

[CONCLUSIONS > FINDINGS] Although the cockpit voice recorder recorded the first officer coughing and clearing her throat numerous times, she stated that she was not sick, and there is no evidence that this (the coughing/clearing her throat) adversely affected the flight or her performance. 3. The accident airplane was properly certificated and maintained and was equipped and dispatched in accordance with applicable regulations and industry practices. There was no evidence of any preexisting powerplant, system, or structural failure. 4. The accident airplane’s cargo and its loading were not factors in the accident. 5. Differences between the MD-11 and MD-10 handling characteristics during the landing phase were not an issue in this accident. 6. Air traffic control was not a factor in the accident. 7. The atmospheric conditions encountered during the approach and landing were within the performance capabilities of the airplane; there was no evidence of significant windshear. 8. The first officer did not properly apply control wheel and rudder inputs to align the airplane with the runway centerline or apply appropriate back pressure on the control column to arrest the airplane’s rate of descent before touchdown; as a result, the airplane touched down extremely hard while still in a crab. 9. The captain, who was conducting a line check of the first officer, did not adequately monitor the first officer’s performance during the final stages of the approach and landing at Memphis and failed to take or initiate corrective action to prevent the accident. 10. The excessive vertical and lateral forces on the right main landing gear during the landing exceeded those that the gear was designed to withstand and resulted in the fracture of the outer cylinder and the collapse of the right main landing gear. Conclusions 58 Aircraft Accident Report 11. A proactive program, similar in concept to FedEx’s enhanced oversight program, in which flight crewmembers who have demonstrated performance deficiencies or experienced training failures are identified and given additional oversight and training would be beneficial to flight safety. 12. The nonrevenue FedEx pilot who opened the L1 emergency exit mistakenly pulled both the manual inflation and slide/raft disengage handles because he was not sufficiently familiar with their operation, thus separating the slide/raft from the L1 doorsill. 13. The guidance contained in the flight crew emergency training section of Federal Aviation Administration Order 8400.10, Air Transportation Aviation Inspector’s Handbook, is not adequate for principal operations inspectors to use in ensuring that emergency exit door/slide training for flight crewmembers is as comprehensive as that which cabin crewmembers receive and is as comprehensive as intended by the regulation. 14. FedEx’s inadequate hands-on emergency procedures training and the differences between the trainer and the door/slide installation on the accident airplane contributed to the unintentional release of the slide/raft. 15. Most of the FedEx pilots on board the accident airplane showed poor judgment and exposed themselves to unnecessary risk when they delayed their evacuation from a burning airplane to salvage personal items. 16.

[ANALYSIS] During postaccident interviews, the captain told investigators that he did not perceive any reason to comment or advise the first officer during that stage of the approach and landing. However, the FDR data (as displayed to investigators through the flight simulator sessions) showed that the first officer neutralized her wind-compensating flight control inputs as the airplane descended through about 100 feet and that the airplane subsequently began to drift to the right of the runway centerline. Although it may have been difficult for the captain, as the nonflying pilot, to immediately recognize that the first officer had neutralized her flight control inputs, the Board’s flight simulation sessions showed that as the airplane’s drift to the right continued uncorrected, the drift became markedly notable from both pilots’ seats as the airplane descended through about 60 feet. The Safety Board understands that a line check airman must balance the need to keep the flight safe with the potential for students to learn from their mistakes; however, in 74 The first officer did move the control column slightly aft immediately before the airplane touched down; however, the input was too late to affect the airplane’s pitch attitude and descent rate before touchdown. Analysis 48 Aircraft Accident Report this case, the captain allowed the airplane to deviate from established and recognizable performance standards at the most critical stage of the approach, without comment or correction. The captain had sufficient time to observe and react to the airplane’s drift and the first officer’s failure to arrest the airplane’s rate of descent as it neared the runway. The investigators (including representatives from FedEx and ALPA) who observed the simulations reported that there were clear indications that aspects of the approach needed correcting and that the captain should have taken corrective actions when these indications became apparent.75 Therefore, the Safety Board concludes that the captain, who was conducting a line check of the first officer, did not adequately monitor the first officer’s performance during the final stages of the approach and landing at MEM and failed to take or initiate corrective action to prevent the accident. 2.3 Failure of the Right Main Landing Gear Assembly The Safety Board’s metallurgical examination of portions of the right main landing gear assembly showed that the landing gear outer cylinder failed in overstress initiating at the shock strut chamber check valve hole surface. The fracture initiated at the base of several circumferential tool marks, which were up to 0.0015 inch deep. The Board’s metallurgical examination indicated that these tool marks were not significant enough to have caused fatigue cracking at the tool mark location during the life of the landing gear.

[ANALYSIS] During postaccident interviews, the captain and first officer described the landing and indicated that it was performed in accordance with FedEx’s crosswind landing procedures. Both pilots stated that the landing flare was normal, including proper alignment with the runway centerline and compensation for wind conditions below 200 feet. The captain indicated that he thought they experienced a “strong gust” of wind during the landing, and both pilots described the landing as firm but otherwise normal. However, FDR evidence and physical evidence, including tire markings on the runway, indicated that the airplane touched down with both main landing gear assemblies and the nose gear well right of the runway centerline on a heading about 5.6° left of the runway heading. The evidence indicated that the airplane’s left main landing gear touched down first about 500 feet from the approach end of the runway, and about 9 feet right of the runway centerline. The evidence further indicated that, almost immediately thereafter, the right main landing gear touched down about 45 feet right of the runway centerline about 49 feet further down the runway. A series of simulations performed using the accident FDR data to backdrive an MD-11/-10 flight simulator indicated that the airplane maintained a stable condition, tracking along the runway centerline with an appropriate wind correction angle as it descended through 200 feet. The FDR evidence further indicated that the first officer began to apply left aileron and right rudder to align the airplane with the runway centerline about 160 feet (this became visually apparent in the simulator about 140 feet). However, the data and the simulations showed that these normal crosswind landing control inputs were only momentary; as the airplane descended below 100 feet, the aileron and rudder control inputs were neutralized and remained neutral until the airplane touched down. The simulations showed that, as a result of the neutralized flight control inputs, the airplane began to drift to the right and continued to drift to the right, with a 5° to 6° left crab angle, until it touched down. The winds at the time of the accident would have required a pilot’s constant attention, significant flight control inputs along both the roll (control wheel) and yaw axes (rudder pedal), and continuous modulation of those inputs. Although the pilots received a windshear alert earlier in the approach, there was no evidence of a significant decrease in wind speed or change in wind direction that might have prompted the first officer to neutralize her crosswind-correcting flight control inputs as the airplane neared the runway. The FDR data and the Safety Board’s performance study showed that after the airplane descended through about 1,500 feet, its pitch attitude, descent rate, and airspeed remained fairly constant until touchdown (about 5° nose up, 720 fpm [about 12.5 fps], and 149 knots, respectively). However, according to FedEx’s crosswind landing procedures, a pilot should begin to increase back pressure on the control column as the airplane descends through about 30 feet to increase pitch attitude and arrest the airplane’s descent rate, dissipate airspeed, and bring the airplane into the correct landing attitude.

[ANALYSIS] The fracture initiated at the base of several circumferential tool marks, which were up to 0.0015 inch deep. The Board’s metallurgical examination indicated that these tool marks were not significant enough to have caused fatigue cracking at the tool mark location during the life of the landing gear. Aside from these tool marks, no significant deviations from drawing requirements were found.76 The MD-10-10F landing gear and associated structure were designed to absorb energy equivalent to a limit vertical speed (descent rate) of 10 fps (about 600 fpm) when landing at the airplane’s maximum design landing weight (consistent with the landing design limits imposed by Sections 25.473 through 25.487). In addition, in accordance with Section 25.723, the MD-10-10F main landing gear is designed to be capable of absorbing reserve energy that is equivalent to a maximum airplane descent rate of 12 fps (about 720 fpm) when landing at the maximum airplane design landing weight. However, during the accident landing, the left main landing gear touched down at a rate of about 12.5 fps (about 750 fpm) and the right main landing gear touched down at a rate of about 14.4 fps (about 864 fpm). As previously mentioned, airplane performance studies conducted by Boeing indicated that the energy dissipated at the right main landing gear during landing was about 34 percent greater than the reserve energy that the landing gear was designed to withstand and about 19 percent greater than the energy dissipated at the left main landing gear.77 In addition, the analyses revealed that peak lateral loads coincided with the excessive vertical loads. Therefore, the Safety Board concludes that the excessive vertical 75 The captain should have verbally prompted flight control actions, commanded a go-around, or taken control of the airplane for a go-around or landing. 76 Nickel deposits were found on the shock strut chamber pressurization/check valve hole surface, which is unusual but not considered contributory to the event. 77 The landing gear external ground loads associated with Boeing’s reserve energy analysis included only vertical and drag loads applied to the landing gear; they did not include the additional effects of lateral or torsional loads that might be imposed on landing gear at touchdown. Analysis 49 Aircraft Accident Report and lateral forces on the right main landing gear during the landing exceeded those that the gear was designed to withstand and resulted in the fracture of the outer cylinder and the collapse of the right main landing gear.

[ANALYSIS] There was no evidence of any preexisting powerplant, system, or structural failure. Cargo loading for the accident flight was routine; no cargo loading anomalies were observed, and the airplane was operating within the prescribed weight and c.g. limits. No hazardous materials were on board the airplane. The accident airplane’s cargo and its loading were not factors in the accident. FedEx records indicated that the flight crewmembers had received the requisite initial, transition, and MD-11/-10 differences training. The Safety Board’s review of the FAA’s FSB report describing its evaluation of MD-11 and MD-10 handling characteristics and numerous postaccident MD-11/-10 flight crew interviews revealed only very subtle differences in flight handling characteristics during the landing phase between the two airplanes. The Safety Board notes that changes in handling characteristics resulting from differences in weight and balance can be much more significant than any differences that could be attributed to MD-11/-10 aerodynamic factors. Further, the significant flight control inputs that are needed when landing either an MD-11 or MD-10 in strong, gusty crosswind conditions (such as those encountered during the accident flight) would render any subtle differences in handling characteristics between the airplanes negligible. In any event, the FDR data show that the first officer momentarily applied flight control inputs to align the airplane with the runway as the airplane descended between about 200 and 160 feet agl; however, she did not maintain those inputs as the airplane neared the runway. In the absence of the necessary and significant flight control inputs during the crosswind landing, any possible differences in handling characteristics between the MD-11 and MD-10 would not be a factor for the pilot. Therefore, the Safety Board concludes that differences between the MD-11 and MD-10 handling characteristics during the landing phase were not an issue in this accident. Analysis 45 Aircraft Accident Report ATC was not a factor in the accident; however, in other circumstances, the ground controller’s delay clearing RMFD ARFF vehicles to the accident site might have had more serious consequences. For additional information, see section 2.6. The pilots were provided with weather forecasts, including AIRMET information, that indicated turbulence below 8,000 feet and strong northerly/northwesterly winds aloft and at the surface. Postaccident pilot and passenger statements and MEM airport surface weather information about the time of the accident indicated that the winds were strong and gusty out of the northwest; however, crew comments recorded by the CVR suggested that the pilots were not overly concerned about the weather conditions. Several airplanes landed without incident in similar conditions during the minutes before and after the accident. The winds identified in the Safety Board’s performance study indicated that the existing weather conditions might have been responsible for the small airspeed gains and losses and the CAWS tail windshear alert experienced by the accident flight; however, there was no evidence of horizontal windshear significant enough to adversely affect the operation of the airplane. Additionally, a meteorological study conducted by MIT at the Board’s request was consistent with the results of the Board’s study.

[ANALYSIS] The airplane was certificated and equipped and maintained in accordance with Federal regulations and approved procedures. There is no evidence that mechanical malfunctions or failures of the airplane structures, flight control systems, or powerplants contributed to the accident. The evidence indicates that the airplane’s hydraulic system was not configured for landing. Because the hydraulic system remained in the low pressure mode, hydraulic pressure was not available to lower the landing gear and deploy the flaps. The flightcrew failed to detect this configuration error and continued its approach into Houston. Comments on the CVR and postaccident statements by the flightcrew indicate that both pilots recognized that the flaps did not deploy after the flaps were selected to 15o, but the flightcrew did not determine the cause of this problem or execute a go-around. The landing checklist was not performed, and the flightcrew did not confirm that the gear was down and locked. The gear warning horn sounded during the approach, indicating that the landing gear was not extended, but it was ignored. When the airplane descended through 500 feet AFE, it was traveling 84 knots faster than the target airspeed of 132 knots. Although, under COA standard operating procedures, this excessive airspeed mandated that the approach be discontinued, the captain rejected a go-around request from the first officer, who was the flying pilot. The GPWS sounded an alert 19 seconds before impact and was ignored. Unaware that the gear was not down, the captain assumed control of the airplane and made a wheels-up landing. This analysis addresses flightcrew performance, including the role of fatigue in the flightcrew’s performance, the adequacy of COA’s oversight of its pilots and the FAA’s oversight of COA, checklist design, and survival factors. 39 2.2 Flightcrew Performance Performance deficiencies exhibited by the flightcrew during this flight include: (1) failure to configure the hydraulic system for landing during the performance of the in-range checklist; (2) failure to detect initially that the flaps did not extend; (3) failure to determine the reason the flaps did not extend after detection; (4) failure to perform the landing checklist and to confirm the landing gear status; and (5) failure to discontinue the approach. 2.2.1 Failure to Properly Complete the In-range Checklist Fifteen minutes before landing, as the airplane descended through 19,000 feet, the captain omitted one item on the in-range checklist. The omitted item, “Hydraulics - ON & HI, CHECKED,” would have enabled the high pressure configuration of the hydraulic system, thereby providing pressure to operate the flaps and landing gear. Three steps were required to complete this checklist item: movement of the AUX and ALT pump switches from “OFF” to “ON,” movement of the left and right engine-driven hydraulic pump switches from “LOW” to “HI,” and confirmation that system pressures were between 2,800 and 3,100 psi.

[CONCLUSIONS > FINDINGS] 3.1 Findings 1. The two-member flightcrew and three flight attendants were trained and qualified to conduct the flight in accordance with Federal regulations. There was no evidence of any medical condition that might have affected the flightcrew’s performance. 2. The air traffic control request to maintain 190 knots to the outer marker did not contribute to the accident because it did not affect crew actions, decisionmaking, or situational awareness. 3. The airplane was certificated and equipped and maintained in accordance with Federal regulations and approved procedures. There is no evidence that mechanical malfunctions or failures of the airplane structures, flight control systems, or powerplants contributed to the accident. 4. Because the captain omitted the “Hydraulics” item on the in-range checklist and the first officer failed to detect the error, hydraulic pressure was not available to lower the landing gear and deploy the flaps. 5. The “Hydraulics” item is placed too low on the in-range checklist, rendering it vulnerable to omission. 6. The captain’s distraction from his duties as pilot-in-command and his disregard for the sterile cockpit rule contributed to the pilots’ failure to detect their hydraulic system configuration error when they selected 5o of flaps. 7. Both the captain and the first officer recognized that the flaps had not extended after the flaps were selected to 15o. 8. The pilots’ lack of previous exposure, either through training or during line operations, to the consequences of improper hydraulic system configuration contributed to their failure to detect their hydraulic system configuration error. 9. The pilots failed to perform the landing checklist and to detect the numerous cues alerting them to the status of the landing gear because of their focus on coping with the flap extension problem and the high level of workload as a result of the rapid sequence of events in the final minute of the flight. 10. Had the landing checklist been properly performed, the flightcrew would have detected the failure of the landing gear to extend. 56 11. Although the first officer was unwilling to overtly challenge the captain’s decision to continue the approach, he did attempt to communicate his concern about the excessive speed of the approach to the captain. 12. There was no compelling reason for the captain’s decision to land the airplane; multiple signals and guidance indicated that the approach should be discontinued, as did Continental Airlines’ standard operating procedures. 13. The flightcrew’s degraded performance is consistent with the effects of fatigue, but there is insufficient information to determine the extent to which it contributed to the accident. 14. There were deficiencies in Continental Airlines’ (COA) oversight of its pilots and the principal operations inspector’s oversight of COA.

[ANALYSIS] It is possible that the horn’s constant tone lost its salience as a signal in the environment because of the extended duration it sounded during the final minute of the approach. However, it is more likely that the pilots failed to detect the numerous cues alerting them to the status of the gear for the same reasons they failed to perform the landing checklist— preoccupation with the flap extension problem and their high workload during the final minute of the flight. The Safety Board concludes that the pilots failed to perform the landing checklist and to detect the numerous cues alerting them to the status of the landing gear as a result of their focus on coping with the flap extension problem and the high level of workload because of the rapid sequence of events in the final minute of the flight. The Safety Board also concludes that had the landing checklist been properly performed, the flightcrew would have detected the failure of the landing gear to extend. 42 The landing gear warning horn stopped for 8 seconds from 0901:07 to 0901:15. This issue is addressed in section 2.2.5.2. 43 See, for example, Getty, D.J., Swets, J.A., Pickett, R.M., & Gonthier, D., 1995, “System Operator Response to Warnings of Danger: A Laboratory Investigation of the Effects of the Predictive Value of a Warning on Human Response Time.” in Journal of Experimental Psychology: Applied, Vol. 1, Pages 19-33. 46 2.2.5 Failure to Discontinue the Approach According to FDR data, 34 seconds before touchdown, the airplane was 504 feet AFE and traveling at 216 knots indicated airspeed. Again, this speed was 84 knots faster than the target airspeed of 132 knots established by the flightcrew during completion of the descent checklist. In addition, the speed was 63 knots faster than the reference airspeed of 153 knots for a flaps-up, slats-extended landing at a weight of 86,000 pounds. The COA DC-9 Flight Manual current at the time of the accident described a stabilized approach as flight on the desired glide path at a steady rate of descent, on the target speed in landing configuration, in trim, and with the proper thrust setting. The manual stated that unstabilized approaches must not be allowed to continue below 500 feet AFE. The approach was clearly unstabilized when the airplane descended through 500 feet; yet, the flightcrew failed to discontinue the approach. 2.2.5.1 Role of the First Officer The first officer told Safety Board investigators that his goal after recognizing that the flaps were not extended was to get the captain to initiate a go-around. Thirty seconds before touchdown, the first officer stated “want to take it around?” and the captain replied “no that’s alright. * keep your speed up here about uh.” When the captain denied the first officer’s request to go around and told him to keep his speed up, the first officer did not challenge the captain’s statement. He also did not question the captain to determine his reason(s) for continuing the approach.

[ANALYSIS] Consistent with other DC-9 pilots who reported failing to properly configure the hydraulic system, this crew detected a problem with flap deployment when the airplane did not respond with pitch and speed changes as flaps were selected to 15o and beyond. However, this crew did not recognize that the failure of the flaps to deploy was a symptom of improper hydraulic system configuration. Neither the captain nor the first officer recalled events concerning improper hydraulic system configuration in his previous DC-9 experience and, therefore, did not 41 At 0900:33, the captain said, “I think the flaps *.” At 0900:35, there were three intermittent sounds from the landing gear warning horn that, according to the first officer, were produced by the captain rapidly moving the throttles back and forth. The throttle manipulation would not have provided the captain with diagnostic information about the flaps. 44 possess firsthand knowledge to help recognize that the symptom he was experiencing was the result of this error. In addition, the Safety Board’s review of the information provided by COA to its pilots concerning the DC-9 hydraulic system revealed that the flight manual and training materials do not explicitly state that if the pumps are not switched to “HI,” the landing gear will not extend and the flaps will not deploy. The Safety Board concludes that the pilots’ lack of previous exposure, either through training or during line operations, to the consequences of improper hydraulic system configuration contributed to their failure to detect their hydraulic system configuration error. The Safety Board believes that the FAA should require all POIs of 14 CFR Part 121 operators using DC-9 and MD-80 airplanes with the “HI, LOW, OFF” hydraulic switch configuration to ensure that operating manuals and training programs include information about the consequences of improper hydraulic system configuration, specifically that the flaps and landing gear will not function normally if the engine-driven hydraulic pumps are not set to “HI.” 2.2.4 Failure to Perform Landing Checklist and Confirm Gear Position In accordance with company procedures, the first officer called for the landing checklist after the gear-down call. Although he placed the gear handle in the down position, the captain never initiated the checklist. Because the flaps remained stowed, the airplane did not slow during the approach. Traveling at a speed of approximately 200 knots, the airplane covered the distance between the outer marker and the runway threshold in about 75 seconds. If the target approach speed of 132 knots had been maintained, it would have taken about 115 seconds to cover this distance. The increase in speed allowed the flightcrew very little time to address the flap problem and configure the airplane for landing. In the 27 seconds that elapsed from the time the captain said “I think the flaps *” to the time the first officer stated “I don’t have flaps,” the captain manipulated the throttles and then responded to the gear-down call, the flaps 25 call, the flaps 40 call, and the flaps 50 call. The captain had very little time to react to the directives he was being given by the first officer.

[ANALYSIS] In the 27 seconds that elapsed from the time the captain said “I think the flaps *” to the time the first officer stated “I don’t have flaps,” the captain manipulated the throttles and then responded to the gear-down call, the flaps 25 call, the flaps 40 call, and the flaps 50 call. The captain had very little time to react to the directives he was being given by the first officer. The first officer’s rapid calls for 40o and then 50o of flaps probably interrupted the captain as he was initiating the landing checklist. It is likely that the first officer was preoccupied flying the approach at an unusually high rate of speed and was unaware that the checklist had not been accomplished. Thus, neither pilot confirmed the landing gear was down and locked as required by the first item in the checklist. During postaccident interviews, both pilots recalled the gear handle being moved to the down position.. Apart from the handle position, the following cues were available to indicate to the flightcrew that the gear was not down: 45 • There was no increase in cockpit noise after the gear handle was placed in the down position, as there would have been if the nosegear doors had opened and the gear extended into the slipstream. • The red gear unsafe lights and the amber gear door open light, which would have illuminated when the gear were in transition, remained off. • The green gear lights, which would have illuminated when the gear were down and locked, remained off. • The gear warning horn sounded almost continuously after the flap handle was moved to 25o at 0900:46.42 Neither pilot was alerted to the status of the gear by the absence of the normal cues (increase in noise and lights). However, detecting the absence of normal cues is often more difficult than detecting the presence of abnormal cues. Nonetheless, neither crewmember responded to the gear warning horn. The reason for this may be that the gear warning horn frequently sounds during routine operations, and it can be perceived by pilots as a nuisance alarm. For example, the horn sounds during approaches whenever the throttles are reduced to idle and the landing gear is not down; a condition that is not always consistent with a dangerous configuration. Research has shown that frequent alarms can lead to slower response times or even disregard for a warning.43 In this case, however, the horn sounded after the gear handle was placed down and the flap handle was moved to 25o. These conditions were outside the traditional “nuisance” envelope. The first officer later stated that he did not hear the horn; the captain stated that he heard the horn but thought it sounded because he put the flaps to 25o before the gear was down and locked. It is possible that the horn’s constant tone lost its salience as a signal in the environment because of the extended duration it sounded during the final minute of the approach. However, it is more likely that the pilots failed to detect the numerous cues alerting them to the status of the gear for the same reasons they failed to perform the landing checklist— preoccupation with the flap extension problem and their high workload during the final minute of the flight.

[CONCLUSIONS > FINDINGS] The flightcrew’s degraded performance is consistent with the effects of fatigue, but there is insufficient information to determine the extent to which it contributed to the accident. 14. There were deficiencies in Continental Airlines’ (COA) oversight of its pilots and the principal operations inspector’s oversight of COA. COA was aware of inconsistencies in flightcrew adherence to standard operating procedures within the airline; however, corrective actions taken before the accident had not resolved this problem. 15. This accident demonstrates the need for all air carriers to bring their checklists that apply to all phases of ground and flight operations into compliance with the contemporary human factors principles of checklist design outlined in the FAA’s report, “Human Performance Considerations in the Use and Design of Aircraft Checklists.” 16. The “C” flight attendant was unable to completely remove the tailcone access plug door, because one of the aft jumpseat shoulder harness straps was buckled to the lap belt, which tied the plug door to the aft cabin bulkhead. Fortunately, the lack of availability of the tailcone exit did not preclude a timely and successful evacuation. 17. Continental Airlines flight attendants received inadequate information and training on the operation of the DC-9 tailcone access plug door. 57 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the captain’s decision to continue the approach contrary to Continental Airlines (COA) standard operating procedures that mandate a go-around when an approach is unstabilized below 500 feet or a ground proximity warning system alert continues below 200 feet above field elevation. The following factors contributed to the accident: (1) the flightcrew’s failure to properly complete the in-range checklist, which resulted in a lack of hydraulic pressure to lower the landing gear and deploy the flaps; (2) the flightcrew’s failure to perform the landing checklist and confirm that the landing gear was extended; (3) the inadequate remedial actions by COA to ensure adherence to standard operating procedures; and (4) the Federal Aviation Administration’s inadequate oversight of COA to ensure adherence to standard operating procedures. 58

[ANALYSIS] Nevertheless, although the lack of spoiler extension alone could account for the captain’s inability to stop the airplane on the ruriway, given the runway condition and the fact that the touchdown was at a point loca’ * over 3,200 feet from the runwe, threshold, the Safety Board examined other factors which also could have adversely affected the airplane stopping capability. These factors relate to the performance of the airplane deceleration devices during a high speed lending and to the evidence of reverted rubber hydroplaning. As shown in figure 9, the almost simultaneous sounds of nose wheel contact, with an increase in the G-trace to about 1.43 G, indicate that the airplane nose wheel and the wheels of one or both main gear struts contacted the runway almost simultaneously. The main gear contact was followed by about 3 seconds of oscillation in vertical acceleration, at slightly less than 1.0 G. Thus, following touchdown, verv little of the airplane weight transferred to tne main gear struts. This situation would have been exacerbated by the captain's stated application of forward pressure on the control column. following touchdown, in order to hold the nose wheel on the runway. Boeing Company data indicate that at an increased airspeed of 15 knots above Vref, the lift generated by PI 487, while in a $-point ground attitude was approximately equal to the eirplane weight. Thus, there would have been little or no weight on the main gear wheels following initial touchdown, a condition which could have been maintained for 4 to 6 seconds with forward displacement of the control column. Since the airplane touched down in heavy rain on a runway that had teen exposed to over 2/3 inch of rain Juring the day, it is possible that initially there would have been insufficient friction between the tires and the runway to obtain a wheel speed of 60 knots or more (the speed required by combination of any two wheels to cause automatic spoiler deployment and automatic movement of the speed brake lever actuator to the "Up" position). Further, the lack of any significant weight on the main gear struts would have prevented the strut compression needed to close the air/ground safety sensor switches which would have precluded the immediate selection of reverse thrust. Tris condition would be consistent with the captain's stated difficulty in deploying reverse thrust. . In acldition, if between the time the captain stated that he attempted first to deploy reverse thrust and then manually deployed the spcilers, or immediately thereafter, he had applied wheel orakes, wheel braking could have begun before the outboard main gear wheels reached synchronous speed with the airplane due to the lack of significant weight on the main gear wheels.

[ANALYSIS] Moreover, the first officer informed the captain at 2006:37 that the speedbrake lever was in manual, i.e., down detent, contrary to Piedmont's requirement that the spced brake lever be armed before landing. The captain's response to that call is unelear on the CVR. ‘Thus, it could not be determined whether he armed the speed brake lever. However, the failure of the ground spoilers to deploy immediately after landing suggests that they were not armed. The GPWS alerted twice thereafter, further indicating that the approach was unstable and not in accordance with company procedures. Since the runway was in sight when the first GPWS alert sounded, and since the first officer called minimums when the second alert sounded, the captain probably recognized that terrain clearance was adequate and, as a result, he believed that he could safely ignore the alert. However, the Safety Board believes that the GPWS was alerting, not because of inadequate terrain elearance, but because of an excessive descent rate close to the ground. Because the airspeed was considerably higher than required at that point and because the airplane had only just been configured for the landing, the captain should have recognized that the approach was not stabilized at the appropriate airspeed, descent rate, and power setting, ‘and consequently, that the margin of safety for landing on a wet ruaway had been reduced to an unacceptably low level. . Because 5 months before the accident Piedmont's flighterew publication, "Operations Update," had discussed the role of proper airspeed management and proper touchdown point to avoid a runway overrun, the Safety Board believes that the captain and the first officer should have been ecutely aware that proper airspeed management was critical. Nonetheless, there {{s no evidence that such airspeed management was present. Rather, the evidence indicates that the airspeed throughout the approach was excessive for the existing runway conditions. As a result, the captain's failure to stabilize the approach compromised his ability to stop the airplane on the runway. Therefore, the Safety Board concludes it was the major factor in establishing the conditions for the accident. . 2.3 Landing and Rollout The evidence indicates that, despite the unstabilized nature of the approach and the touchdown that was at a point considerably beyond the recommended touchdown point, on a runway that contained areas of standing water, the airplane could have been stopped on the remaining runway had the captain made optimal use of the airplane decelerative devices, I.e., spoilers, thrust reversers, brakes and antiskid system. FE We NL RENE SY a be Ge oe ep pder toc Sek ae ES leon ec eee py ad.

[FINDINGS] Therefore, the Safety Board urges the FAA to issue an operations bulletin to principal operations inspectors of air carriers operating aircraft with flight attendants informing them of the need to cease providing alcohol to passengers who are In, or appear that they are about io be In, an intoxicated state. 3. CONCLUSIONS Findings 1. The flighterew and the flight attendants were properly certificated and qualified for the flight. 2. The airplane was properly maintained for the flight. 127 Aircraft Accident Report--"Simmons Airlines Flight 1746, Embraer Bandeirante, EMB-110P1, Near Alpena, Michigan, March 13, 1986" (NTSB/AAR-87/02). cia keh me HNN ne Ney i RES RT RON RETR RP Canter ON Et es ARR NTN A MORE UREN Marg © om iy seroma NL ah f Me te A mm = ag ae nomen Om A Ran yatta ae mm 37~ Air traffic control services provided to this flight were in accordance with acceptable procedures. Weather factors did not contribute to the aceident. There was no evidence of preexisting damage to the airplane structure, systems, or powerplants that could have contributed to the accident. The airplane was not configured for landing until just before touchdown, contrary to Piedmont operating procedures. The GPWS alert just before touchdown indicated an excessive rate of descent. The approach was flown contrary to Piedmont operating procedures. ‘he captain should have elected to discontinue the approach because it was not carried out in accordance with Piedmont operating procedures and because the airplane was not configured for landing until just before touchdown. Crew coordination was deficient due to the first officer's failure to call the captain's attention to aspects of the approach that were not in accordance with Piedraont operating procedures. The airplane touched down over 3,200 feet from the approach end of the runway, #t an alrspeed that was excessive for the prevailing runway surface conditions. The spoilers were not deployed {{iinmediately after teuchdown which adversely affected the airplane stopping performance. ae i e The captain probably applied wheel brakes prematurely after touchdown which may have resulted in the loss of brake effectiveness on the outboard wheels. The concrete culvert located beyond the departure end of the runway caused most of the damage to the airplane. The friction on runway 36R was generally acceptable; however, in the last 1,500 feet, it was unacceptable and this contributed to the severity of the accident. The airplane hydroplaned during the substantial portions of the last 1,500 feet of roll on runway 36R. The evacuation was effective and completed within 1 1/2 minutes. PERERA ts Sosa ar SSE SN oe ee The emergency response to the accident was deficient in the limited number of ambulances dispatched to the site. Two passengers were reported to have been intoxicated at the time of the accident, and they could have adversely affected the evacuation. TREE RES, SET WS Sie tlc manle ee bron one

[ANALYSIS] However, the evidence suggested by the position of the cockpit controls as to whether the spoilers had deployed is inconclusive for several reasons. Damage to the underside of the airplane precluded a determination of the amount of right main gear strut compression needed to operate the air/ground safety sensor switches and to open the ground spoiler interlock valve. Consequently, the rigging tolerance of 11/2 inch plus 11/2 inch minus 0 inch for the interlock valve, and the rigging tolerance of 5 inches for the air/ground safety sensor could not be verified, and it could not be conclusively determined as to whether the spoilers functioned as designed. I I The damage to the airplane underside also prevented a determination of possible actions of the speed brake electric actuator electrical circuits. Ordinarily, once extended, the actuator will not retract simply by moving the speed brake lever to the "Down" detent. The thrust levers in the forward thrust regime must also be advanced. However, if a wheel rotational speed of 60 knots or more was never achieved on any combination of two main gear wheels throughout the landing roll, the spoilers would not have automatically deployed and the actuator, automatically extended. In addition, during the erash sequence, an action in the actuator's eleetrival circuits could have caused the actuator to retract. Thus, the retracted position of the actuator was inconclusive as to its relationship to spoiler deployment. Sines it is likely that a wheel rotational speed of 60 knots or more on any - combination of two main gear wheels was obtained at some point during the landing rol}}, the possibility that the speed brake lever was never moved from the "Down" detent was examined to determine the effect on the airplane stopping performance. According to data supplied by Boeing, on a wet, grooved runway without _ Spoiler extension, the airplane could have been stopped in about 3,500 feet provided full reverse thrust was obtalned within 3 seconds of touchdown. However, significant reverse thrust was not obtained until about 12 seconds after touchdown when the airplane had slowed to about 112 KIAS, at which time less tian 1,900 feet of runway remained. (See figure 8.) Consequently, although the frictional qualities of the last 1,500 feet of DT we nega TEE TRE on Ree Tee ? + <' 2 . _ . + La See Leese pe teg ey eit Lote (esnjert SSS ae Sepa NES am, ONT - a - mse inb drag RIS ett sneasae nme” Ear he ange SBE CNET E to pae eee nL AMEE runway 36R were substandard, the lack of any spoiler extension could account for the delay in obtaining reverse thrust, the airplane poor stopping: performanze with the last 1,900 feet of runway, and its departure trom the runway at 4 speed in excess of J KIAS.

[ANALYSIS] Tris condition would be consistent with the captain's stated difficulty in deploying reverse thrust. . In acldition, if between the time the captain stated that he attempted first to deploy reverse thrust and then manually deployed the spcilers, or immediately thereafter, he had applied wheel orakes, wheel braking could have begun before the outboard main gear wheels reached synchronous speed with the airplane due to the lack of significant weight on the main gear wheels. As demonstrated in FAA-sponsored research with a Lockheed L-1011 and a Boeing 737, achieving a main gear wheel spin up that is synchronous with the airplane can take as long asi 2 seconds, following a normal landing at Vref on a smooth, wet runway. Consequently, even with satisfactory transverse grooving in the area of the runway wheve PI 467 touched down, the combination of heavy rain and minimal wheel loading could have delayed wheel spin up extensively. . During the 6 seconds after touchdown, the airplane slowed from about 147 knots to about 135 knots, a deceleration of about 2 ft/sec2. Since the brakes on the inborrd wheels would not have been available immediately due to locked wheel protection, the performance of PI467 in the first 6 seconds after touchdown suggests that the outboard wheels may have locked up within 3 to 4 seconds after touchdown, most likely due to the: captain's application of wheel brakes before synchronous speed was achieved. ae SEAR ES RaT BEARER Meet eats he leah eee eate Vo SAT EADS ESRI IS el Baa Reale Wn 5 apt AYR He SY es ICN AQ UEP I Se ROS AEE OO MEY Aap cn ce Serer ELE TPAC pet ND ABEL HSiED at nieebrne eh een pniahd ar orb kin tReet scalar tateg eco PUERTO MSE Mer npn Nt ember eR aR en FRE MOR Ei SE GR LI ON BOTY LUNE REL CL SNOT ROE a Sah is AR ab eater NOY TN TERING Cea NN TEL EN IR SETS % : ; During that 6~seeond period, the airplane rolled about 1,400 feet. As the NASA research with pressure-blas-modulated antiskid systems demonstrated, on an ungrooved, wet runway, brake application before synchronous speed is reached can result in a locked wheel condition. This can prolong the time required for the wheels to reach synchronous speed by as much as 25 seconds. Under locked wheel Conditions, the effective braking coefficients are less than 0.05, at speeds over 70 knots. At 2007:24, the airplane began a rapid deceleration compared to a previous 6-second period of slight acceleration. Its airspeed decreased about 20 knots from 135 to 115 seconds, for a deceleration rate of about 8.5 ft/sec2. Since significant reverse thrust was not generated until about 2007:31, during the 4-second period from 2007:24 to 2007:28, sufficient weight was probably transferred to the main struts, apparently by extension of the flight spoilers, to allow the air/ground safety sensor to sense ground operation.

[ANALYSIS] In summary, given the many variables that affected PI467's stopping performance on runway 36R, the Safety Board could not determine conclusively whether or not the spoilers were extended following touchdown. However, irrespective of whether the spoilers were extended, the excessive speed of the airplane as it entered the last 1,500 feet of runway led to the hydroplaning that precluded effective braking action. Consequently, the Safety Board concludes that the accident was directly reiatcd to the manner in which the captain flew the approach and executed the landing. 2.4 Crew Coordination The Safety Board believes that, while the decision to continue the approach belonged to the captain only, the first officer participated In the decision-making process in the Information he provided the captain. The first officer recited the landing checklist: and stated that the speed brakes were in the manual mode of operation. He also called out the approach lights wher they became visible. Sh Weanenmann: vebvanen area sist bey se saaaimitamenateuaimts wumenieielitlalel srernve menenma aie prongs eeanauere nace top srerepaeametinenitaiehinmenieien RRR neh: ae Bf! : . A eae ttteck RS: ia see wats = Saaty MT Te eT eae ota se ep tt és ere <f J 6 tye ESE Te NC Fre ree ay Py eae a! 73 Fam rata 2 ~32~ The first officer's statement about the speed brake lever being in manual, contained the clear implication that it was not armed as required. ‘This was, the Safety Board believes, a subtle re:rainder to the captain that the required approach and landing procedures were not being adhered to. At the same time, the first officer did not point out to the captain that the airplane was still net configured for landing when it was well inside the final approach fix, and he did not call out to the captain that the airspeed was excessive throughout the approach. Therefore, the Safety Board concludes that the first officer's lack of assertiveness in providing the captain with needed information and the captain's failure to respond to the "subtle" callout of the speed brakes in manual are indicative of deficient crew coordination, also known as cockpit resource management, and that this deficiency contributed to the accident. . The Safety Beard is aware of the difficulty that first officers face in attempting to provide captains with needed information at critical points in a flight, when I such attempts could be distracting. More important, perhaps, is the difficulty they may face when attempting to influence the pilot-in-command to reconsider and possibly alter a decision. Thus, it would have been very difficult, onee inside the final approach fix, for the first officer to suggest to the captain that the approach was not stabilized and, as a result, they should go around. Such a suggestion could, if presented inappropriately, distract the captain and could potentially endanger the safety of flight.

[ANALYSIS] This would have permitted deployment of the thrust reverser deflectors and the application of the inboard. wheel brakes as well as deployment of the ground spoiler panels. However, by the time significant reverse thrust was generated, as evidenced by the engine sounds recorded on the CVR, less than 1,900 feet of runway remained. Since the last 1,500 feet of runway 36R contained a crown that was insufficient to promote adequate drainage, collapsed and uneven transverse grooving, and less than retommended friction qualities, the 1,900 feet of runway remaining was insufficient to stop the alrplane from a speed of 112 KIAS, even with all decelerative devices operating. The actual performance of PI 467 during the last 12 seconds indicates that comparatively little deceleration (3.33 ft/sec2) was obtained. The airplane left. the runway at a speed of abut 72 KIAS. It is also apparent from the skid marks on the last 100 feet of runway 36R and from the condition of the four main gear tires that the airplane experienced reverted rubber hydroplaning before it left the runway. To achieve reverted rubber hydroplaning, wheel/tire rotational speed must be reduced essentially to zero so that the tires skid along the runway surface. Since there was no evidence of preexisting defects In the antiskid braking system, wheel/tire rotational speed could be reduced by one of two ways}} either by disengaging the antiskid system or by having the rotational speed already reduced to zero before rubber reversion took place. Sinee there was no indication that the captain disengaged the antiskid system, the Safety Board believes that wheel/tire rotational speed was reduced essentially to zero by a combination of dynamie and viscous hydroplaning that preceded the reverted rubber hydroplaning. Further, the evidence indicates that the poor frictional qualities of the last 1,500 feet of runway 36R and the pooled water on the runway surface contributed to the dynamic and viscous hydroplaning. In summary, given the many variables that affected PI467's stopping performance on runway 36R, the Safety Board could not determine conclusively whether or not the spoilers were extended following touchdown. However, irrespective of whether the spoilers were extended, the excessive speed of the airplane as it entered the last 1,500 feet of runway led to the hydroplaning that precluded effective braking action.

[ANALYSIS] Piedmont procedures specified that before crossing the LOM the final landing flap setting should have been selected and the airspeed should have been reduced to a level appropriate for that flap setting. On this flight, the final flap setting was 30° and the final approach airspeed or Vref was 131 knots. The CVR indicates that the final flap setting was not accomplished until the airplane was cr the glide slope, well inside the final spproach fix. Further, the first officer did not lower the gear until 2005:39, and the captain did not select the final 30° flap setting until 2006:48, when the airplane was less than 1 mile from the runway threshold and 2 seconds before the first officer made the 500 feet (ag!) call. Moreover, the airspeed was not reduced to 131 knots until after landing. Thus, the approach of PI 487 was carried out in a manner well outside the parameters established in Piedmont's procedures. wees ee RANT ain terme lly pe a The Safety Board believes that because the airplane was not configured for the landing until 500 feet above touchdown, the captain was "behind" the airplane. That is, he was setting flaps, lowerirg the landing gear, and trying to reduce the airspeed after the flight was descending on the glide slope and well inside the final approach fix. Had the captain slowed the airplane and configured it as required before reaching HA ‘YOU, he could have stabilized the approach and controlled the airspeed with the needed precision. . ‘Instead, the airplane crossed HAYOU at 194 KIAS, crossed the threshold about 165 KIAS, and touched down about 147 KIAS, considerably higher than the Vref speed of 131 KIAS, over 3,200 feet from the runway threshold and over 2,000 feet beyond the company recommended touchdown point. Under these circumstances, the margin of safety was reduced considerably, that is, the captain's ability to stop the airplane on the remaining “¥ runway depended on his ability to optimally use the airplane decelerative devices, with no ‘ s margin for error allowed in the use of those devices. fo og Oe Ls Te rere Sy Tee e eS em ‘ i Ace ARORA In EN aS he RI REN RRR RT SME RAY RAE Se et Despite the captain's assertions that he added 20 knots to Vref because of his concern for a wind shear condition, the Safety Board believes that, if correct, he failed to al properly interpret and apply guidance provided on the subject in the company operations i 7 manual. (See appendix D.) From that guidance, with surface wind reports, the lack of a significant convective activity, and his knowledge of the tailwind on the approach, the ; 0° 4 captain should have known that the existing wind shear involved that of a tail wind a shearing to a light crosswind or no wind. Under these conditions, significant speed additions are not needed and may compound airplane controllability because this type of PN wind shear tends to increase indicated airspeed during descent, through the reducing tallwind shear.

[ANALYSIS] FE We NL RENE SY a be Ge oe ep pder toc Sek ae ES leon ec eee py ad. Fee ee aie) Live adsorb etuen dwt gs Sl we Eu maga tach oon Sve MART OES Mane be ne eet ua oe nd ge en ae ete yt Sag hatin ata ies. .. 2d + 5 eee erg PE eek Oo SS oa od, 7 . wen caw vere ewer seen “+ fev ert We EL mR AY ar Me EL sat be, Beas eer re aenP were any LO atcha nly Katine ee tes ae ATS a Ga ek aa rE NC CLT TST SAA EBA LR OTA GEE 2 COREE RID DEAT BEEN I eT RTA SR ot ~27- However, the evidence suggests that, despite repeated guidance in the Piedmont operations manual on the need to arm spoilers and, if not armed, deploy them upon touchdown before the other decelerative devices, the spoilers were not armed and were not deployed. This can be readily accounted for by the rushed nature with which the approach was conducted and the extent to which required procedures were not followed, both on the approach and upon touchdown, as well as by witness stateménts and supportive evidence from the Piedmont B-737 simulator. However, the captain stated that he did arm the speedbrakes before landing but that they failed to deploy autornatically. As a result, the Safety Board closely examined the airplane performance following touchdown on tunway 36R to determine the consequences to its performance following speedbrake arming, given the environmental and airplane conditions at the time. The captain stated that immediately following touchdown, he attempted to deploy the thrust reversers, without success. He said that he then moved the speed brake lever to the "Up" position to manually deploy the spoilers, and then immediately applied the wheel brakes. After the airplane left the runway, according to the captain, he configured it for an evacuation, in complia :e with emergency procedures, by retracting the spoilers and by moving the speed brake lever to the Down" detent. After the accident, the speed brake lever was found in the "Down" detent and the speed brake lever actuator was found in the retracted position. However, the evidence suggested by the position of the cockpit controls as to whether the spoilers had deployed is inconclusive for several reasons. Damage to the underside of the airplane precluded a determination of the amount of right main gear strut compression needed to operate the air/ground safety sensor switches and to open the ground spoiler interlock valve.

[(a) FINDINGS] Only because of alert, responsive, and orderly conduct of these military passengers, many of whom took charge during the enerrency, was an even greater disaster averted. 2.2 Conclusions (a) Findings 1. The aircraft was certificated and maintained in accordance with existing regulations. The pilots were certificated and qualified for the flight. The aircraft was within certified weight and balance limitations for the takeoff. The aircraft rolled into position on Runway 6R and held for approximately 1 minute 30 seconds before the takeoff was initiated, A thin layer of ice covered the runway surface. A braking torque of unknown source was imparted to all eight main landing gear wheels. The main landing gear wheels did not rotate during the attempted takeoff. The fact that the initial sliding coefficient of friction on the runway surface was only slightly higher than the normal rolling coefficient of friction of the wheels masked the detection of the locked wheels, because of the frictional drag created by the rubber degradation, tire failure, and abrasive milling of wheel rims, the acceleration was adversely affected snd the eireraft did not attain the necessary lift-off speed. The slower than normal. acceleration of the aircraft wus nol evident to the pilots until such time that a successful rejected takeoff was virtuwlly impossible. - 28 . 1. The impact conditions were classified as survivable with ull fatalities resulting from the post-impact fire. 12, Some flight attendants were incapacitated as a result of body restraint system, and galley equipment security dericlencies. Their incapacitation precluded their effective assistance in passenger evacuation. (vb) Probuble Cause The National Transportation Safety Board determines that the probuble cause of this accident was the fuilure of the aircraft to attain the necessary airspeed to effect Lift-off during the attemptcd takeoff, The lack of acceleration, undetected by the crew until after the aircraft reached V) speed, was the result of a high frictional drag which was cuused by a failure of all main landing sear wheels to rotate. Although it was determined that a braking pressure sufficient to lock all of the wheels was imparted to the brake system, the source of this pressure could not be determinec. Possible sources of the unwanted braking pressure were either a hydraulic/brake system malfunction or an inadvertently enraged parking, brake,

[ANALYSIS AND CONCLUSIONS] The above comparison confinas the coefficient of friction tests applicable to the initial phase of the takeoff wherein the aircraft perfornmance up to a speed of approximately 100 KIAS vas just slightly below the normal exoucted performance. Thus, detection by the crew that the wheels were not rotating and the attendant progressive performance degradation would have been difficult, if not impossible, during the early stages of tre takeoff. Perhaps the only cue could have been an vnusua? feel of the aircraft at the initial breakaway. This thought was regated by the crew in their statements that the sensation cf brake release was felt at the outset of the takeoff run. Fron the foregoing discussion tt is obvious, then, thet the primary causal area concerns the reas‘.:, o2 reasons the main landing gear wheels failed to rotate during the takeoff. The possibilities for this unwanted condition are many, however, the evidence available in this case clearly indicates that a sustained braking torque, which was somehow applied to all of the main landing gear wheels subsequent to alignment on the runway, prevented any further rotation of them. There was no evidence found, or supportive data developed, which would indicate tit a phenomenon such as hydroplaning had inhibited the wheel rotation. In considering the conditicns under which an equal braking torque, sufficient to lock all wheels, could have been applied, the following possibilities were raised: - A malfunction occurred in the brake system or hydraulic system which either applied an unwanted brake pressure or prevented complete release of the brakes. High frictional forces developed by improperly installed wheels created sufficient resistance so as to prevent wheel rotation. - 22. - The brakes were applied by the crew while in position on the runway and were unintentionally not released prior to the takeoff attempt. Extensive examination of the brake assemblies revealed no indications of any faiiure or malfunction to these components. The parking brake mechanism was intact and operational and was in the released position. All clearances between the brake plates were normal and the discs showed no evidence of overheat, binding, welding, or any other abnomality that could have been associated with a braking torque problem. The air brake lever was found in the "Off" and safetied position evidencing that no intentional application of the air oreke occurred. Because the air brake cylinder was not recovered there was no way of determining if there had been an inadvertent application of air to this system which activated the brakes. However, this possibility is also rather remote in that a leaking air valve is designed to vent overboard and not into the System, thereby preventing the application of brakes. The possibility of a malfunction within the hydraulic system leading to an unwanted brake application was also examined. Various system failure mode conditions were postulated and exemined as to their effect on the brake system. It was found that under certain, albeit remote, conditions a flow of hydraulic fluid in excess of nornal quantity could raise the pressure on the brake supply lines, through the return system, and apply brakes.

[ANALYSIS AND CONCLUSIONS] As was noted, evidence of skidding in the direction of takeoff was observed at each of the four tire prints made by the left truck. Skid marks from the right-hand inboard truck were observed just a short distance from the left gear static footprint. Frogres:ive deterioration of all main lending gear tires began at the initiation cf the takeoff and continued the entire length of the runway. The first scrap of reverted rubber was located only 560 feet from the start of the takeoff and by 2,700 feet from the starting point, the amount of fiber in the rubber scraps indicates that some or all of the tires were grown! down to their carcass reinforcing cords. It was determined that by 4,300 feet fron the start of the takeoff, all of the left-hand tires were flat and by 8,700 feet all of the righthana ties were flat. Examination of the tires and wheels which were not extensively firedamaged revealed that all were ground down in one contact area only, with no evidence to suggest that they had ever rotated during the attempted takeoff. The type of tire damage and blowout patterns appeared typical of that caused by locked-wheel skids. X-ray examination of all tires, excent the No. 8 tire which was destroyed by fire, showed that none of the tires had rolled after it had gone flat. In view of the above, it is concluded by the Board that all of the main landing gear wheels of Ni909C rolled as the aircraft was taxied onto the runway and that they never rolled thereafter. The crew stated that the initial acceleration or movement of the aircraft appeared quite normal following the application of takeoff power and brake release. The reason the crew did not detect the fact that the initial movement of the aircraft was a skid becomes easily comprehensible if considered in terms of the NASA runway friction data. Assuring a total weight on the landing gear of approximately 349,000 pounds and a breakaway ccefficient of friction of 0.14, only 48,900 peunds of friction drag could be created. With a total engine thrust at 1.86 EPR (Nk900C's takeoff EPR) equal to 74,600 pounds, only 65 percent thrust would have been required to cause the atrecraft to skid even with brakes on and wheeis locked. Since the sliding coefficient of friction (0.025) ie almost a full order of magnitude Jower than the breakaway coefficient of friction (0.14), a surge of acceleration pcssibly similar to a normal takeoff brake release would have been felt when the aircraft first started to move.

[PROBABLE CAUSE] PROBABLE CAUSE The National Transportation Safety Board determines that the probable cause of this accident was the failure of the aircraft to attain the necessary airspeed to effect lift-off during the attempted takeoff, The lack of acceleration, undetected by the crew until after the aircraft reached V1 speed, was the result of a high frictional drag which was caused by a failure of all main landing gear wheels to rotate, Although it was determined that a braking pressure sufficient to lack 81) of the wheels was imparted to the vrake system, the source of this pressure could not be deternnined. Possible sources of the unwanted braking pressure were either hydraulic/brake System malfunction or an inadvertently engaged parking brake. RECOMMENDATIONS AS a result of this investigation, the Safety Board recommended that the Federal Aviation Administration take the following actions: (a) Detennine and implement takeoff procedures that will provide the flightcrew with time or distance reference to appraise the aircraft's acceleration to the V, speed. (b) Initiate action to incorporate in its airworthiness requirements, a provision for fuel System fire safety devices which ‘vill be effective in the prevention and control of both in-flight and post-crash fuel System fires and explosions. The Board further recommends that the Federal Aviation Administra-~ tion in cooperation with the aircraft manufacturers and the Nattonal Aeronautics and Space Agency, utilize the results of extensive research and accident investigetion data to Gevelop and implement major improvements in the design of transport aircraft interiors. Of particular concern are the crashworthiness of galley equipment stewardess seats and restraining devices, and the flenmability of cabin interior materials, = 3 = 1. INVESTIGATION 1.1 History of Flight Capitol International Airways, Inc., Flight C203/26, a DC-8-63F, N&gOSC, was a Military Airlift Command (MAC) contract flight scheduled from McChord Air Force Base, Tacoma, Washington, to Cam Ranh Bay, Republic of Soutn Viet Nam, with en route refueling stops at Anchorage, Alaska, and Yokota, Japan. The flight departed from McChord AFB at 1204 1/ on November 27, 1970, with 219 passengers and a crew of 10 aboard. It landed on Runway 6L at Anchorage International Airport at 1532. There were no wusual occurrences en route and the flight was described by the crew as routine. The captain stated that during the landing rollout he used reverse thrust and medium heavy braking to bring the aircraft to a stop on the icy runway. Braking action was fair to poor and only licht braking was used while taxiing to the ramp.

[ANALYSIS AND CONCLUSIONS] All of the wheel bearings were in operational condition and there were no unusual surface markings or discolorations to indicate high frictional activity. Similarly, the bearing cups were in good order and showed no evidence of secring or overheating. Under the category of an unintentional and unwanted braxe application, consideration was given to the possitility of an inadvertent foot pressure on the brake pedals during the takeoff by either the captain or first officer. The captain stated that he held the brakes with his instep on the rudder bar and his toes on the brake pedals while the engine power was being stabilized. Then, simultaneous with the throttle advance to takeoff power he released the pressur? on the brake pedals keeping his feet on the rudder pedals. ‘rhe first officer stated that during the takeoff his feet were placed on the rudder pedals with his heels on the floor and that all steering was accomplished in this manner. He stated that he did not feel the brake pedals being depressed at any time during the takeoff. With the existing slippery conditions of the runway and corresponding sliding coefficient of friction, only slight braking pressures would have been required to allow the aircraft to begin its initial slide from the takeoff position and to continue to the point where catastrophic degradation of the tires was in effect. However, when the uircruft began to slide the rise in the coefficlient of friction most certainly would have been sufficient to overcome dragging brakes, if in fact, the cause of the condition was due to an inadvertent and slight braking pressure being applied to the pedals by one of the crewmembers. In that cuse, some indication of wheel rotation would have been evidenced ecither on the tires or the runway. In addition ~ ak - to the fact that no such evidence was found’, it is also difficult to believe that the brakes could be applied and maintained equally in this manner without a consclous effort on the vilot's part to do sc, It is, therefore, highly improbable that this possibility 5 was res ponsible for the locked wheels. j ' ; ; i The remaining possibility involves an unreinembered act on the part of the crew, of setting the parking brakes while holding on the runway awaiting takeoff clearance and then failing to release the brakes prior to commencing the takeoff. Notwithstanding the fact that both the captain and first officer tertified that the parking “rakes were not applied at any time subsequent to departure from the terminal ramp, it is known that this type of situation his happened in the past and, therefore, the possibllity of.a similar occurrence in this case was closely analyzed by the Board. In most cases where flighterews have overlooked checklist items, or have failed to configure an aircraft properly for a particular flight regime, one of two factors, or u combination thereof, have intervened to cause a memory lapse. These factors are a time interval between actions activities, and an occurrence of a significant distraction prior to the required function.

[ANALYSIS AND CONCLUSIONS] Some of these cases involved all of the brakes and others involved one main landing gear only. -19. 2. ANALYSIS AND CONCLUSIONS Ze” Analysis The evidence developed during the investigation of this accident showed that the nain landing gear wheels were not rutating during the takeoff run. As a result, the aircraft, operating within 988 pounds of its maximum structurai weight limit cf 350,000 pounds, failed to attain the computed lift-off speed of 163 KIAS. The entire usable Length of Runway 6R, which was coated with ice, had been used in attaining the highest speed recorded of 152 knots. Considerable testing and analytical studies were conducted to determine the cause of the locked wheels as well as the operational consequences relating to the performance of the aircraft, It was noted that after tie aircraft taxied into position on Runway 6R, it remained there.for approximately 1 minute and 30 seconds before the takeoff was ccumenced. This position on the runway was marked by a static footprint cr the left main landing gear tires. These tires, which left clear tracks from the taxiway onto the runway, appeared to have rolled into the position marked by the static footprint, and, as evidenced by skid murks on the runway, apparently all four of these tires skidded out of that position. The static rootprint 3. 3 caused whe the heat of the tires melted through tne ice covering on the runway. The heat necessary to melt the Jce was most likely generated as a result of the long taxi run from the teiminal to the runway (approximately 2 miles) at a very heavy gross weight. According to one study concerning heat generation for rolling tires, taxiing 1 mile at this aircraft's gross weight would have heated the air inside the tire to 160° F, It then follows, that a 2-mile taxi run would heat the tires to an even greater degree and, considering the time that the aircraft was in position on the runway, they would have meited through the ice as exhibited by the footprint. The Board is unable to determine why there was no footprint from the right main landing gear. However, tt is possible that the ice on the runway was not of uniform thickness so that there was little or no ice on the mmnway surface under the right main landing gear. As was noted, evidence of skidding in the direction of takeoff was observed at each of the four tire prints made by the left truck. Skid marks from the right-hand inboard truck were observed just a short distance from the left gear static footprint. Frogres:ive deterioration of all main lending gear tires began at the initiation cf the takeoff and continued the entire length of the runway.

[CONCLUSIONS > FINDINGS] The NTSB concludes that a lightweight recording system conforming to EUROCAE ED-155 would have helped determine the flight crew’s actions during the landing and subsequent go-around attempt, including, but not limited to, whether they silently conducted checklists (partially or completely), which flap settings they selected, and how much braking effort they made upon landing. 86 NTSB Aircraft Accident Report 3. Conclusions 3.1 Findings 1. The investigation found that the pilots were properly certificated and qualified under federal regulations. 2. The investigation found that the accident airplane was properly certificated, equipped, and maintained in accordance with federal regulations. Examinations of the recovered components revealed no evidence of any preimpact structural, engine, or system failures. The airplane was within normal weight and balance limitations. 3. The accident was not survivable. 4. The captain allowed an atmosphere in the cockpit that did not comply with well-designed procedures intended to minimize operational errors, including sterile cockpit adherence, and this atmosphere permitted inadequate briefing of the approach and monitoring of the current weather conditions, including the wind information on the cockpit instruments; inappropriate conversation; nonstandard terminology; and a lack of checklist discipline throughout the descent and approach phases of the flight. 5. The flight crewmembers exhibited poor aeronautical decision-making and managed their resources poorly, which prevented them from recognizing and fully evaluating alternatives to landing on a wet runway in changing weather conditions, eroded the safety margins provided by the checklists, and degraded the pilots’ attention, thus increasing the risk of an accident. 6. The airplane touched down within the target touchdown zone and at the recommended touchdown speed, and the captain likely applied sufficient pressure on the brakes during the initial part of the landing roll to take full advantage of the available runway friction, but he failed to immediately deploy the lift-dump system after touchdown in accordance with company procedures, which negatively affected the airplane’s deceleration. 7. No evidence exists that reverted rubber or dynamic hydroplaning occurred. 8. If the captain had continued the landing and accepted the possibility of overrunning the runway instead of attempting to execute a go-around late in the landing roll, the accident most likely would have been prevented or the severity reduced because the airplane would have come to rest within the runway safety area. 9. Establishing a committed-to-stop point in the landing sequence beyond which a go-around should not be attempted for turbine-powered aircraft would eliminate ambiguity for pilots making decisions during time-critical events. 10. If, as a 14 Code of Federal Regulations Part 135 operator, East Coast Jets had been required to develop standard operating procedures and its pilots had been required to adhere to them, many of the deficiencies demonstrated by the pilots during the accident flight (for example, 87 NTSB Aircraft Accident Report inadequate checklist discipline and failure to conduct an approach briefing) might have been corrected by the resultant stricter cockpit discipline. 11.

[ANALYSIS] The flaps were found in the wreckage set to the fully retracted position (flaps 0°), which was an incorrect setting for a go-around and would have made it more difficult for the airplane to lift off, and this setting may reflect the confusion and lack of crew coordination that can follow from unprofessional compliance with use of nonstandard terminology. NTSB Aircraft Accident Report 63 landing roll. The NTSB notes that other recent overrun accidents have not been as catastrophic because the flight crews did not attempt to go around after landing.114 The NTSB has previously investigated accidents during which the pilots did not commit to the landings and made a delayed decision to go around. For example, on October 5, 2005, a Beechcraft 58 overran the runway in Jacksonville, Florida, after attempting a go-around late in the landing roll on a wet, ungrooved runway. During postaccident interviews, the pilot stated that the airplane touched down on the first quarter of the runway at about 100 knots. He stated that, past the midfield point, “the airplane still had a lot more momentum to bleed off,” so, with only one quarter of the runway left, he attempted a go-around. He stated that, when he noticed that the airplane was not climbing, he aborted the go-around and overran the departure end of the runway. In addition, on July 15, 2005, a Cessna 525A collided with a localizer antenna in Newnan, Georgia, after the pilot conducted a go-around late in the landing roll on a wet, ungrooved runway. The pilot stated that he applied brakes upon landing and that the airplane then hydroplaned. He stated that he chose to abort the landing with 2,300 feet of runway remaining (the runway was 5,500 feet long). As a result of the pilot’s delayed decision to go around, the airplane became airborne only 300 feet from the runway end. Both of these accidents might have been prevented if the pilots had committed to the landings or better understood where the committed-to-stop point was rather than attempting to go around with insufficient runway available to lift off and clear obstacles.115 The NTSB concludes that establishing a committed-to-stop point in the landing sequence beyond which a go-around should not be attempted for turbine-powered aircraft would eliminate ambiguity for pilots making decisions during time-critical events.

[ANALYSIS] The NTSB notes that because the runway was ungrooved and wet, the friction it could provide would have been lower than that achievable on a dry runway, and it would not have required maximum braking effort to achieve the maximum available braking forces on the tires. An increase in brake pressure beyond that required to achieve maximum available braking forces would only tend to lock the wheels, and the brake anti-skid system would then release brake pressure to avoid this situation. The CVR recorded the sound of the airplane touching down at 0945:04 and, at 0945:07, a sound consistent with the airbrake handle moving to the OPEN position. One second later, the first officer stated, “we’re dumped,” and, immediately after, “we’re not dumped.” In response to the first officer’s last statement, the captain stated, “no, we’re not,” while simultaneously making straining sounds, consistent with physically attempting to move a cockpit control. The CVR then recorded a sound consistent with the airbrake handle moving into the DUMP position. According to the airplane performance study, the airplane’s airspeed over the threshold was the reference 60 NTSB Aircraft Accident Report 61 landing airspeed of about 122 knots, and the airplane touched down about 1,128 feet from the runway threshold, which is within the target touchdown zone.112 East Coast Jets procedures call for the immediate deployment of full lift dump upon touchdown. However, the CVR evidence indicates that upon touchdown the captain only moved the airbrake handle to the OPEN position instead of fully aft to the DUMP position and likely did not fully deploy the lift-dump system (full flaps and airbrake deflection) until about 7 seconds after touchdown, which was not in accordance with company procedures (or the deceleration device deployment times used to develop the BAe 125-800A AFM wet runway guidance material). ( See section 2.6.) The captain should have deployed lift dump by moving the airbrake handle in one motion to the DUMP position, not partially deploying the airbrakes and then fully deploying lift dump. The first officer most likely stated, “we’re dumped,” as an automatic callout upon landing when he saw the captain move the airbrake handle aft. The latter callout, “we’re not dumped,” likely resulted from the first officer’s required check of the flap position indicator and provides an example of effective monitoring by the first officer. The NTSB concludes that the airplane touched down within the target touchdown zone and at the recommended touchdown speed and that the captain likely applied sufficient pressure on the brakes during the initial part of the landing roll to take full advantage of the available runway friction, but he failed to immediately deploy the lift-dump system after touchdown in accordance with company procedures, which negatively affected the airplane’s deceleration. None of the tires exhibited flat spots or evidence of reverted rubber, and all of the tire tread depths were within specified limits. Therefore, the investigation ruled out the occurrence of reverted rubber hydroplaning.

[ANALYSIS] None of the tires exhibited flat spots or evidence of reverted rubber, and all of the tire tread depths were within specified limits. Therefore, the investigation ruled out the occurrence of reverted rubber hydroplaning. The NASA friction expert reported that the runway was generally in “excellent” condition because it had a relatively high macrotexture, consistently adequate cross-slope throughout the runway length, insignificant rubber deposits, and no concrete surface deterioration. Further, pavement drainage models indicated that the runway was fully capable of draining the rainfall reported throughout the morning of the accident; therefore, there was not sufficient standing water on the runway at the time of the accident to have caused the airplane to experience dynamic hydroplaning. The NTSB concludes that no evidence exists that reverted rubber or dynamic hydroplaning occurred. 2.2.3 Captain’s Decision and Subsequent Attempt to Go Around The pilots remained silent for 10.5 seconds after the captain stated, “no we’re not” (acknowledging that the airbrake handle was not in the DUMP position before moving it to that position), until the captain called out “flaps” at 0945:22. About the same time (more than 17 seconds after touchdown), the CVR recorded a sound consistent with increasing engine noise and the initiation of a go-around. The results of the airplane performance study indicated that, at the time that the go-around was initiated, the deceleration rate was such that the airplane would have exited the runway end at a ground speed of between 23 and 37 knots and stopped between 112 According to the pilot/controller glossary in the AIM, the touchdown zone is “the first 3,000 feet of the runway beginning at the threshold.” NTSB Aircraft Accident Report 62 100 and 300 feet into the 1,000-foot-long runway safety area. Therefore, it can be reasonably assumed that, at some point during the landing roll, the captain likely became concerned that the airplane would run off the runway end and had to decide whether it was preferable to overrun the runway or attempt a go-around. However, as discussed, no evidence indicates that the captain was prepared for the possibility of a go-around. Specifically, he did not conduct an approach briefing, which would have included briefing a missed approach. It is possible that the captain’s decision to go around was delayed because it took time for him to realize that the airplane was not decelerating as he expected and the possibility of a runway overrun was increasing. In addition, he might have been waiting, expecting the airplane’s deceleration to improve.

[ANALYSIS] The effective braking coefficients determined in the airplane performance study for the accident landing were substantially below those defined by the BCAR RWHS and AMJ 25X1591, assumed in the CAPS simulation, and underlying the wet runway landing distances provided in the BAe 125-800A AFM guidance material. The study also determined that the braking coefficients most representative of the actual performance of the airplane during the landing closely matched those calculated by a NASA friction expert using the CFME measurements made on the runway 2 days after the accident. The performance study also indicated that the total landing distances computed using the Section 25.109 braking coefficients 82 NTSB Aircraft Accident Report 83 can be significantly longer than those computed using the AMJ 25X1591 coefficients and provided in the AFM. For example, using the accident landing weight of 19,912 pounds, no wind, the OWA field elevation, the outside air temperature on the day of the accident, 140-psi tire pressure, and deceleration device deployment times, the airplane performance study indicated that the total landing distance using the AMJ 25X1591 braking coefficients was 3,338 feet and that the total landing distance using the Section 25.109 braking coefficients was 4,225 feet, which is 26 percent longer.137 (See table 4 in section 1.16.3.4.2) Assuming an 8-knot tailwind, the total landing distances were 3,792 and 4,928 feet, which is 32 percent longer, respectively. Adding the 15-percent safety margin recommended by SAFO 06012 to these distances, the required runway lengths with an 8-knot tailwind would be 4,361 feet using AMJ 25X1591 data and 5,667 feet using Section 25.109 data. Therefore, even if the accident flight crew had conducted an arrival landing distance assessment using the existing AMJ 25X1591-based AFM data (for either wind condition), it would have shown that the airplane could have stopped on the 5,500-foot runway with a safety margin of more than 15 percent. As shown, a landing distance assessment using Section 25.109 data would have indicated that the runway length was insufficient for landing with at least a 15-percent safety margin with an 8-knot tailwind. The airplane performance study indicated that the airplane would have exited the runway at between 23 and 37 knots and stopped between 100 and 300 feet beyond the runway end, but within the 1,000-foot runway safety area. The Section 25.109 calculations are consistent with current knowledge about wet runway braking performance, which is reflected in the engineering data used by the FAA to update regulations governing the calculation of accelerate-stop distances for wet, ungrooved runways138 and by the TALPA ARC in drafting new recommendations to require and support arrival landing distance assessments.

[CONCLUSIONS > FINDINGS] Both pilots’ performance was likely impaired by fatigue that resulted from their significant acute sleep loss, early start time, and possible untreated sleep disorders, and fatigue might have especially degraded the captain’s performance and decision-making abilities when he had to decide while under time pressure whether to continue the landing or initiate a goaround. 19. Although the first officer took a prescription sleep aid for which he did not have a prescription the night before the accident, because of the short duration of its effects for most individuals, it is unlikely that the use of this medication degraded the first officer’s 88 NTSB Aircraft Accident Report performance at the time of the accident, which occurred about 12 hours after he took the medication. 20. Allowing civil aviation pilots who have occasional insomnia to use prescription sleep medications that have been proven safe and effective would improve these pilots’ sleep quality and operational abilities. 21. Educating and training pilots on fatigue-related issues could prevent pilots from operating flights while impaired by fatigue. 22. Formal guidance on how pilots can be treated for common sleep disorders while retaining their medical certification could help mitigate fatigue-related accidents and incidents. 23. The wet runway landing distances provided in aircraft flight manuals or performance supplemental materials that are based on the braking coefficients defined by the British Civil Air Regulations Reference Wet Hard Surface and Advisory Material Joint 25X1591 can be significantly shorter than the actual distances required to stop on some wet, ungrooved runways. 24. Title 14 Code of Federal Regulations Part 135 pilot-in-command line-check requirements are not adequate because they allow more than one required inspection to be conducted simultaneously and do not require that the line checks be conducted on flights that truly represent typical revenue operations; thus, the efficacy of line checks to promote and enhance safety is minimized, and pilots have limited opportunities to demonstrate their ability to manage weather information, checklist execution, sterile cockpit adherence, and other variables that might affect revenue flights. 25. Although the enhanced ground proximity warning system terrain database had not been updated to the most current version, the outdated database was not a factor in the accident. 26. A lightweight recording system conforming to European Organization for Civil Aviation Equipment ED-155, “Minimum Operational Performance Specification for Lightweight Flight Recorder Systems,” would have helped determine the flight crew’s actions during the landing and subsequent go-around attempt, including, but not limited to, whether they silently conducted checklists (partially or completely), which flap settings they selected, and how much braking effort they made upon landing. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the captain’s decision to attempt a go-around late in the landing roll with insufficient runway remaining.

[FINDINGS] The fatig e crack in the skin developed over a period of at least 2 flight-hours before the failure of the tail rotor blade. The fatigue crac.. in the spar developed over a period significantly longer than 2 hours. The fatigue crack in the skin of the tail rotor blade was probably present when the blade wa: inspected 45 min, before the accident. However, the crack could nat be ~_*tected readily. The visuai inspection procedures in effect were not adequate to detect the fatigue crack in the skin which developed from a fatigue crack in the spar. The loss of the 35-in. section of the ta:l rotor blade did not cause a loss of directional control. The loss of the 35-in. section of the tail rotor blade generated unbalanced forces, which caused three of the four tail rotor gearbox housing attachment flanges to break from static overload and a portion of a thread of the bolt in fourth fiange to fail. ~24- The failure of the ettachiment flanges permitted the gearbox and assemby to separate from the tail pylon, which caused the loss of directional control and an abrupt nosedown pitch attitude change. The helicopter's low altitude when directional control was lost and the nosedown pitch change made further contrel of the helicopter impossible. The flightcrew's decision to attempt an emergency landing at Newark International Airport was an appropriate decision under tte circumstances, 18. The accident was partially survivable. 19. The failure of the seat support and tiedown structure contributed to the number of fatalities and serious in!:-cies. 20. Fewer serious injuries would have occurred if the passengers had taken a "brace" position. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the separation of the tail rotor assemb:y and gearbox from the aircraft at an altitude which made further controlled flight impossible. The rotor assembly and gearbox seperated because of severe vibrations in the rotor assembly which were induced by the loss of a tail rotor blade due to fatigue failure. Contributing to the severity of the passengers’ injuries were the seat feilures wriech occurred when the deceleration forces exceeded the relatively low design strength of the FAA-approved seats, and the lack of guidance on a passenger brace position for emergency landings.

[ANALYSIS] The Safety Board could not determine if the bolt had partially backed out before the failure of the tail rotor tlade, or as a result of the failure of the blade. Any adverse effects of a partially backed out attachment bolt would probubly be shared equally by all five tail rotor blades. However, the investigation did not indicate any additional fatigue failures in the remaining blades. The Sarety Joard coneludes that a loose tail rotor gearbox wes not the cause of the fatigue initiation, Flighterew Decisionmaking The flightcrew was certificated properly und was qualified for the flight. They hed received the off-duty time required by regulation, and there was no evidence thet medical or physiclogieal factors might have affected their performances, When the tail rotor failed, Flight 972 was at an altitude of about 1,400 ft, ana 1 mile east of the airport over the congested Port Newark area. The crew recognized inmediately that the tai rotcr system had malfunctioned. As a result, they had two opt.ons- - either to return to the airport or to autorotate. They chose to atternpt an emergency landing ut the airport for several reasons; (1) The airport, which was the only suitable fcrced-landing srea, was iess than a mile away. (2) Once the appropriate emergency checklist had been accomplished and the helicopter had been slowed to between 60 and 70 kns, the ere’. was able to control the helicopter in spite of the severe vibrations. (3) There: was no yaw to indicate a drive shaft malfunction and resultant loss of tail rotor rpm, (4) They knew that if autorotation was initiated the rotor rpm would increase to about 105 to 108 percent, which would place a greater load on the tail rotor system. (5) The firs\ officer believed tha, the additional rpm would have increased the vibration and .aused complete luss of control of ine helicopter. (6) There was no suitable landing «cea below them. Autcrotation may have resulted in an earlier landinz, but it may also have caused the gearbox to separate sooner because of increased rotor rpm's and increased stress on the gearbox mouniing flanges. Also, it probably would have resulted in a landing in a hazardous area of Port Newark. ‘fherefore, becuse extended flight was not neressary, the helicopter was not yawing: significantly, and the flichterew had no reason to believe fail in the short ;eriod of time needed ‘to reach the ‘airport, the Safety Roard concludes that the flighterew's decision to niake a controlled emergeney landing at autorutational speed was an appropriate decision When the tail rotor gearbox and assembly seprrated from the tail pylon. che loss of the 180-Ib unit caused tne center of gravity to change immediately ‘o 260 in., or within 2 in. of the forward limit, and resulted in an abrupt nosedowr change in pitch.

[ANALYSIS] Extensive «‘dewall flexion was probuble in the forward fuselage of Flight $72 because the landing gear lateral supports were attached to the outside cf the fuse'age in this area, The forces on the landing gear would have been transmitted directly to the fuselage walls and flooring, causing localized flexion of the walls end distortion of the floor. As a result, the passengers in the furward cabin were probably thrown when the sidewall tiedown structure failed. The Safety Board is aware that similar inadequacies in design requirements for passenger seats for general av-stion aircraft e.ust, Therefore, improvements in general ay.ation aircraft crashworthiness have been made a special safety objective. We expect to include improvements in design requirements for helicopter passenger sets in this objective. Four ctcupants reported taking a brace position before the initial impact. These ce.upants receive? less severe upper torso and head injuries than other occupants seated in the same rows who did not assume a orace position. For example. one passenger whe was in the brace postioi received only minima! injuries although both cecupants seated beside him were killed. Another passenger, a former U.S. Army helicopcer crewman, assumed a brace position and received only minimal tad and upper torse injuries while the passenger next to him received a serious open, depressed fronta! skull fracture. The emergency procedures and the pessenger briefing cards should have specifically required the flight attendant to instruet passengers to assume the standard brace position, which would have red.ced tue possibility of serious injuries during the emergency landing. ~23- 3. CONCLUSIONS Findings 1. 2. The flighterew was certificated properly and qualified for the flight. The aircraft was certificated, maintained, and dispatched in accordance with Federal Aviation Regulstions and approved maintenance procedures. The black tail rotor blade broke when the helicopter was at 1,200 ft and abeut 1 mile east. of the airport. The blade broke because it had been weakener Sy & preexisting fatigue erack in the leading edge spar and the blade skin about 35 in. from the tip of the blude. The facigue crack had propagated through 90 percent of the leading edge spar and through less than 2 in. of the skin covering the spar. The spar fatigue origin was located at the outboard corner radius of the aft face of the leading edge spar. The area where ‘he fatigue originated in the spar was covered by the aluminum envelupe, which made it impossible to detect during a visual inspection. The fatigue crack began in the spar before it started in the skin. The fatig e crack in the skin developed over a period of at least 2 flight-hours before the failure of the tail rotor blade. The fatigue crac.. in the spar developed over a period significantly longer than 2 hours. The fatigue crack in the skin of the tail rotor blade was probably present when the blade wa: inspected 45 min, before the accident. However, the crack could nat be ~_*tected readily.

[ANALYSIS] Simultaneously, directional control was lost when the tail rotor fell off. The helicopter began an immeciate descending rignt turn of about 270° before hitting the ground. Although 4!! directional contrcl was lost, had the gireraft been at a higher a'ticude, it might have been fl:red sufficiently to cushion the touchdown on the airport; however, touchdown still would have been accomplished from a spiraling turn. Although the first officer did attempt to slow the rate of descent by initiating a flare before impact, the nosedown attitude und the low altitude of the helicopter prevented him from raising the nose high enough to cushion the toucndown, Sur vivabisity The acvident was partially survivable because the g forces were within the range of human toierance and because of the minimal disruption of the fuselage structure and miriimal reduction in the oeeupiable volume of the fuselage. (Sce Figure 10.) The fatalities and severe injuries were caused by failures of passenger restraint systems under comparatively high vertical g forces in the forward portion of the helicopter. The vertical g loads, which exceeded the 4-g minimum certification standard imposed for helicopter seats by Civil Acronauties Manual 7.260, were variable, but they probably were in the range of 15 g's in the forward cabin. This estimate is based on the damage to the helicopte:. In the 15-g range of impact foree, the U 5. Army Crash Survival _ design Guide, TR 71-22, October 1971, indicates that only moderate injuries should ve expected when adequately restrained persons encounter impact forces of this magnitude. The impact forces caused the sidewall tiedowns to fail and separate, which ecuused many of the seats to separate or become loose. The sidewall tiedowns probably separated when the fuselage sidewalls flexed on impact. Controlled erash testing of various aircraft and helicopters conducted by the National Aeronautics and Space Administration indicate that fuselage sidewalls flex several inches in crashes similar to that experienced by Flight 972. As a result, seat pan attachmerts will separate from the fuselage. Once the seat pan has separated, it ~Pi- will swing through a wide are until the passengers come into contact with other objects. Extensive «‘dewall flexion was probuble in the forward fuselage of Flight $72 because the landing gear lateral supports were attached to the outside cf the fuse'age in this area, The forces on the landing gear would have been transmitted directly to the fuselage walls and flooring, causing localized flexion of the walls end distortion of the floor.

[ANALYSIS] Initially, the spar crack may have propagated as a result of lower frequency stress cycles, such as startup/shutdown loads, or when the blade was exposed te higher stresses during maneuvers. In this case, the amount of flight time for total crack extension would be significantly more than 2 hours, The provisions of the various Service Bulletins had been adhered to, and al! of the required inspections had been made and documented by New York Airways, Inc. However, since the section of the spar which failed was completely enclosed in an aluminum envelope, a fatigue crack which had begun in the spar could not be seen during a vis'‘al inspection and would not necessarily distort the skin of the blade at a specified .. ae before failure. The examination of the failed blade incicated that the fatigue crack in the skin was present when the second inspection was made, about 45 minutes before the accident. At that time the erack was less inan 2 in. long. Since a typical fatigue crack is tight, the crack in the skin of the blade was probably difficult to see, especially since it was less than % in. long. Consequently, while inspection procedures were adequate for the detection of certain faults in a tail rotor biade, they did not assure detecticn of a skin fatigue crack which resulted from the fatigue failure of the spar. ~19- Failure of the Tail Rotor Gearbox and Assemply The failure of the rotor blade generated an unoalanced foree in the rotor system which caused three of the four tait gearbox attachment iugs to rail under static overload, As a result, the tail rotor and tah rotor gearoox assemblies separated from the helicopter and complete directional control was lost. The fourth tail gearbox attechment lug remained intact and in place on the tail gearbox housing. The lug hole and the corresponding hole in the year bos attachinent plate were elongated. The bo}}t was bent slightly, and the threads damaged badly. The barrel nut for the bol¢ was in place in the pylon and relatively undamagea., she lack of damage to the threads of the barrel nut indicated that the bolt had lost torque and partially backea out of the nut before the tail rotor gearbox and assembly separated from the tail pylon. The damage to the threads of the left aft bolt couid only have occurred if the bolt had at least partially backed out and if the pearbox was capable of motion relative to the mounting plate at the bolt hole. Relative motion between these components could occur only if another of the gearbox attach points had been broken. Therefore, tne leit aft bolt threads were damaged after the black tail rotor blade separated and at least one of the gearbox attachment lugs had been fractured. The Safety Board could not determine if the bolt had partially backed out before the failure of the tail rotor tlade, or as a result of the failure of the blade. Any adverse effects of a partially backed out attachment bolt would probubly be shared equally by all five tail rotor blades. However, the investigation did not indicate any additional fatigue failures in the remaining blades.

[ANALYSIS] Variation of the wind direction would result in a change in the location of updrafts and downdrafts in relation to the ridge. In addition, the presence of a horizontal axis vortex 36The unbroadcast ATIS wind direction, velocity and gust factor would have been similar to that of the earlier ATIS broadcast. 78 on the lee side of the ridge could have produced a localized updraft along the flightpath of the airplane. The Safety Board concludes that, although the variable wind conditions at the time of the accident may have caused localized updrafts and downdrafts in the area, the DFDR data indicates that there were no large-scale updrafts or downdrafts that would have affected the accident aircraft. During the approach to runway 15, to the point at which flight 1572 struck the trees, the airplane would have encountered moderate turbulence and localized updrafts and downdrafts due to the interaction between strong low altitude winds and rough terrain along the flightpath. Windshear due to strong gusty low altitude winds also occurred following the tree strike, as the airplane was on approach to the runway. An estimated mean wind profile indicated a decreasing headwind as the airplane descended to the runway. Although windshear was occurring as the airplane approached and passed over the ridge line, it was the gustiness of the low altitude winds, rather than a small-scale weather feature, that significantly affected airplane performance. Airspeed excursions amounted to only about 10 knots. Further, a descent rate of about 1,100 feet per minute was initiated by the flightcrew from about 1,840 feet msl and was maintained until tree contact. The linear nature of the pressure altitude trace indicates that the airplane’s flightpath was probably not significantly affected by updrafts, downdrafts, or windshear. Such an effect would be seen as a deviation from the near linear pressure altitude trace. Therefore, the Safety Board concludes that the decreasing headwind shear seen in the estimated mean wind profile data was not severe enough to cause the flightcrew to deviate below the MDA. Along the approach to runway 15, cloud bases were near 2,000 feet with multiple cloud layers above, and the tops of the clouds were above 15,000 feet. Flight visibility was near 0 miles in the clouds and 2 to 3 statute miles below the lowest cloud base. The first officer reported "there's the runway straight ahead " at 0055:57. The airplane was about 2.1 miles from the end of the runway at this time. Moderate rain probably occurred along the approach to the runway with more intense rain near the runway. Given the above described conditions, the Safety Board concludes that the weather at the time of the accident was not severe enough 79 to cause the aircraft to deviate below the MDA, and did not contribute to the accident. 2.8.1 LLWAS Equipment The northwest LLWAS sensor was physically out of alignment by 38 degrees and was corrected subsequent to the accident.

[CONCLUSIONS] CONCLUSIONS FIndingS 0... cece ceecceescceesseceseceseececseecceseecseseecesaeceeeeeceseeceeaeeeeseeeeseseneeeengs Probable Cause .o........c ce ceceeeeececcccccccccssseseececcceccccansaccececceceeeeunaasseseeceececs RECOMMENDATIONS 00 ccceeeceseceseeeeeceesaeceeceaeesaeessaeesees APPENDIXES Appendix A--Investigation and Hearing ..0..... ec eeeceesecesteeeeecestecenteeenees Appendix B--Cockpit Voice Recorder Transcript ...........cc:eeeseeeseeeteees Appendix C--Excerpt from AAL Flight Manual on Altimeter Setting Procedure.........ceeecsccccessceseseeesseeceeeeeesseecesecseseeeeueceseeeeeeeenseees vi EXECUTIVE SUMMARY On November 12, 1995, at 0055 eastern standard time a McDonnell Douglas MD-83, N566AA, owned by American Airlines and operated as flight 1572, was substantially damaged when it impacted trees in East Granby, Connecticut, while on approach to runway 15 at Bradley International Airport (BDL), Windsor Locks, Connecticut. The airplane also impacted an instrument landing system antenna as it landed short of the runway on grassy, even terrain. Flight 1572 was being conducted under Title 14 Code of Federal Regulations, Part 121, as a scheduled passenger flight from Chicago, Illinois, to Bradley International Airport. The National Transportation Safety Board determines that the probable cause of this accident was the flightcrew’s failure to maintain the required minimum descent altitude until the required visual references identifiable with the runway were in sight. Contributing factors were the failure of the BDL approach controller to furnish the flightcrew with a current altimeter setting, and the flightcrew’s failure to ask for a more current setting. The safety issues in the report focused on tower shutdown procedures, non-precision approach flight procedures, precipitous terrain and obstruction identification during approach design, the issuance of altimeter settings by air traffic control, low level windshear alert system maintenance and recertification, and emergency evacuation issues. Recommendations concerning these issues were made to the Federal Aviation Administration.

[CONCLUSIONS > FINDINGS] At the time of the accident, the indicated altitude (height above airport elevation) that the airplane’s QFE altimeter was indicating was about 76 feet too high (based on the altimeter setting received at 0030), resulting in the airplane being 76 feet lower than indicated on the primary altimeters. 13. Although the flightcrew did not use the most current QNH setting they had available (29.40 inches of Hg.) in the standby altimeter, this error did not affect the accident sequence of events because the flightcrew had the correct, but outdated, QFE setting (29.23 inches Hg.) in the altimeters they were using when the accident occurred. 87 14. If the first officer had monitored the approach on instruments until reaching minimum descent altitude (MDA) and delayed his search for the airport until after reaching the MDA, he would have been better able to notice and immediately call the captain’s attention to the altitude deviation below the MDA. 15. The excellent crew resource management and flight skills that the flightcrew used, as reflected on the CVR recording following their encounter with the trees, were directly responsible for limiting the number of injured passengers to one individual. 16. FAA quality control was inadequate for accurately resolving the height of the trees on the ridge line. 17. There is great value in flying non-precision approaches with a constant rate or angle of descent until the airport environment can be visually acquired, if the avionics aboard the airplane can safely support such a procedure. 18. The FAA should have, but did not, consider the issue of precipitous terrain when developing and modifying the approach to runway 15. 19. The entire ridge line on the final approach course to runway 15 at BDL is an obstacle and it, and similar terrain close to other airports, should be fully depicted upon the appropriate approach charts. 20. Considering the fact that the pressure changes were described by the weather observer as "pressure falling rapidly," and especially in light of the controller’s failure to issue the current altimeter setting (29.38 inches Hg.) upon initial radio contact, and his 0044:34 entry of 29.34 inches Hg. in the ARTS system while the accident aircraft was on his frequency, it would have been prudent for the approach controller to have issued the 88 altimeter setting changes as the airplane neared the airport. 21. Closure of the tower was a good managerial decision because the safety of people in the tower was compromised by the adverse wind and rain. 22. The TRACON supervisor’s communications with the flight were appropriate and aided the flightcrew. He acted in a professional manner and should be commended for his willingness to assist the flight under the circumstances. 23.

[ANALYSIS] The Safety Board concludes that the flightcrew’s failure to maintain the required MDA until the required visual references identified with the runway were in sight directly caused this accident. 68 2.5 Actions After Tree Strike Regardless of the fact that the flightcrew descended below the MDA, their actions after the initial tree strike were noteworthy. The leading edge and trailing edge flaps of the airplane were severely damaged, power from one engine failed almost immediately after the tree strike, and power from the other engine failed shortly thereafter. Any mistakes in glidepath management, ground track management, or airplane configuration timing by the captain and first officer would probably have caused the airplane to land in unsuitable terrain prior to the clear area at the end of the runway and undoubtedly would have resulted in more severe injuries to crew and passengers. The Safety Board concludes that the excellent crew resource management and flight skills that the flightcrew used, as reflected on the CVR recording following their encounter with the trees, were directly responsible for limiting the number of injured passengers to one individual. 2.6 Terminal Instrument Procedures (TERPS) As discussed below, the manner in which AVN-100 procedures specialists evaluated the obstacles on the instrument approach in question is markedly different from that of the flight procedures inspectors in the FIAO, making it possible to come to a different conclusion concerning the height of obstacles along the flightpath. In spite of the procedures that required the FAA FIAO to coordinate with the flight procedures specialists in the event of data or charting errors, such coordination apparently was never effectively accomplished. The instrument procedures development program was separated from the flight operations inspection program in 1994. The procedures specialists, who design the instrument approaches, are now part of the FAA’s Office of Aviation System Standards (AVN) and design an approach based upon charting methodology, rather than actual physical surveys to determine obstacle clearance surfaces. The specialists never directly measure obstacle heights, glidepath angles, and other variables, when they design an approach, but rather they rely upon graphs, charts, maps, and tables of information to do so. During the initial development of the BDL VOR runway 15 approach, the procedures specialist that designed the approach determined, based upon charts, that a 55 foot obstruction existed within the required obstacle clearance plane of the VASI. Further, if a VDP were to be established at the 69 intersection of the VASI with the MDA, the same obstruction would penetrate the required obstacle clearance plane by 55 feet. This was inconsistent with the FIAO determination, both during the approach development and after the accident by means of flight inspection, that the obstacle clearance plane was not penetrated by the ridge line and trees.

[CONCLUSIONS] The safety issues in the report focused on tower shutdown procedures, non-precision approach flight procedures, precipitous terrain and obstruction identification during approach design, the issuance of altimeter settings by air traffic control, low level windshear alert system maintenance and recertification, and emergency evacuation issues. Recommendations concerning these issues were made to the Federal Aviation Administration. NATIONAL TRANSPORTATION SAFETY BOARD WASHINGTON, D.C. 20594 AIRCRAFT ACCIDENT REPORT COLLISION WITH TREES ON FINAL APPROACH AMERICAN AIRLINES FLIGHT 1572 McDONNELL DOUGLAS MD-83, N566AA EAST GRANBY, CONNECTICUT NOVEMBER 12, 1995 1. FACTUAL INFORMATION 1.1 History of Flight On November 12, 1995, at 0055 eastern standard time (EST)1, a McDonnell Douglas MD-83, N566AA, owned by American Airlines (AAL) and operated as flight 1572, was substantially damaged when it impacted trees in East Granby, Connecticut, while on approach to runway 15 at Bradley International Airport (BDL), Windsor Locks, Connecticut. The airplane also impacted an instrument landing system (ILS) antenna as it landed short of the runway on grassy, even terrain. Flight 1572 was being conducted under Title 14 Code of Federal Regulations (CFR), Part 121, as a scheduled passenger flight from Chicago, Illinois, to Bradley International Airport. On November 10, 1995, at around 1700, the captain, first officer and three flight attendants reported to Washington’s National Airport (DCA) to begin a 3-day flight sequence together. The scheduled departure time was at 1800 and consisted of three flight segments the first day from DCA to Nashville, Tennessee, (BNA) continuing to Chicago’s O’Hare International Airport (ORD) and to Denver, Colorado (DEN). The airplane departed DCA at 1758, according to company records, flew the three segments, and arrived at DEN at 0310. The crew was on duty for 10 hours and 25 minutes, and had accumulated 5 hours and 53 minutes actual flight time. Due to the crew’s late arrival in DEN, the regularly scheduled layover of 15 hours and 18 minutes was reduced to 13 hours and 35 minutes. 1All times in this report are eastern standard time unless otherwise noted. 2 At 1615, on November 11, the flightcrew departed the hotel in DEN for a scheduled check-in time of 1700. The airplane they were to fly arrived late from ORD; the inbound flightcrew reported that N566AA was a “good airplane.” Flight 1572 originated in DEN and departed there at 1809, with the first officer as the flying pilot, arriving at ORD at 2047. The flight was 23 minutes late, based on the scheduled arrival time of 2024. The flightcrew stated that the airplane performed normally and that the flight was uneventful.

[ANALYSIS] Because the Safety Board is concerned that non-precision approaches at airports other than BDL may be adversely affected by wind and turbulence associated with precipitous terrain, the Safety Board believes that the FAA should review and evaluate the appropriateness of the let-down altitudes for all non-precision approaches that have significant terrain features along the approach course between the initial approach fix and the runway. Airline safety departments and pilot labor organizations, such as the Allied Pilots Association and the Air Line Pilots Association, should be consulted as part of this review. In addition, the Safety Board believes that the FAA should solicit and record user comments about difficulties encountered in flying a particular approach to evaluate approach design more accurately. 2.6.2 Approach Plate Terrain Depictions The single 819-foot obstacle depicted on the final approach course of most BDL runway 15 VOR approach plates could lead flightcrews to believe that there was one discrete obstacle, and that it was the only dangerous point on the final approach (see Figure 1). However, the Safety Board concludes that the entire ridge line is an obstacle, and that it and similar terrain close to other airports should be fully depicted on the appropriate approach charts. As an example, see Figure 4, the BDL approach plate used by British Airways. The Safety Board continues to believe, as reflected in Safety Recommendation A-96-102, following the accident near Buga, Colombia, that the FAA should require that all approach and navigation charts graphically present terrain information. 72 (BRADLEY INTL) WINDSOR LOCKS VOR or TACAN 15 fenecs SSA 25m 48 Popaa I IRight onto BDL 149A to BOC Ee I [2830 te DODAYAEDL 114 & 2000 I 1830 I FAF to MAP (THR) 4.8nm 4693062 1, Final approach track 4608 loft of extended centrale, O00 from threshold. 2, Tranginon: AGL to MISTR vee Tra26M. 1Onm. Man alt, 3500, 3.1700 trem DAUUN eampate af a 3.5" angle of deacam, 4. 0ME indicates 02a at thraahald. Figure 4.--BDL approach plate used by British Airways. 73 2.7 Air Traffic Control Factors 2.7.1 ATC Altimeter Setting Distribution Procedures The approach controller is required to issue the QNH (above sea level) altimeter setting on initial contact with an arriving flight, in accordance with the Air Traffic Control Handbook, FAA Order 7110.65. AAL flight 1572 first contacted the approach controller at 0043:41. The controller should have issued the current altimeter setting of 29.38 inches Hg. at that time. The controller said that the omission was inadvertent.

[ANALYSIS] The Safety Board concludes that the accident approach was not stabilized as the airplane descended through 500 feet agl and that the pilots should have detected this and performed a go-around. As the pilots continued the approach, the flight engineer read the before landing checklist query items, with the captain responding, in accordance with FedEx procedures. About 0536:49, the CVR recorded the first officer commenting that he was going to have to stay a little higher or he was “gonna lose the end of the runway.” The airplane’s subsequent descent rate gradually decreased to about 960 fpm; however, the airplane remained significantly below the proper glidepath. FDR data showed that, shortly thereafter, the engine power began to increase further, reaching about 1.34 EPR about 0537:12. During this time, the captain stated that the runway was starting to “disappear…a little” but then said, “think we’ll be alright.” 106 As previously noted, the captain’s and first officer’s VSIs were not instantaneous—that is, the instruments displayed data that lagged behind the airplane’s actual rate of descent, which was derived from the FDR data. However, the manufacturer’s simulations indicated that the pilots’ VSIs would have displayed a rate of descent of at least 1,000 fpm when the airplane descended through 500 feet agl. Analysis 55 Aircraft Accident Report In postaccident statements, the flight crew and ground observers indicated that there were no obstructions to visibility along the approach path. However, the comments made by the first officer (“gonna lose the end of the runway”) and captain (“disappear…a little”) suggest that they may have encountered a temporary obstruction to visibility (for example, clouds or mist) as they approached runway 9. If such an obstruction existed, it may also have obscured the PAPI lights. Although a temporary obstruction might help explain the flight crew’s failure to recognize the PAPI guidance while that obstruction was present, it does not explain why the three pilots failed to recognize the presence of four red PAPI lights throughout the rest of the approach. Further, according to FedEx procedures (and FAA regulations), if the approach end of the runway became obscured at any time during the visual approach, the pilots should have performed a go-around. The night visual approach to runway 9 at TLH was conducted over a protected national forest area that was devoid of ground lights or other visible references by which the pilots could judge their height above terrain. FedEx’s recurrent training module on black hole approaches, which the first officer received in 1999,107 warned that pilots conducting visual approaches at night over terrain with minimal visible ground features or lighting often perceive the airplane to be higher than its actual altitude. Research has shown that in situations like this, a pilot typically flies a lower-than-normal approach until the error starts to become apparent (usually about 2 to 3 miles from the runway), at which point the pilot takes corrective action.

[ANALYSIS] The airport lighting systems, including the PAPI lights, were not a factor in the accident. 105 A pilot might observe a pink PAPI signal as a result of condensation on the PAPI lenses or briefly when the airplane is transiting the narrow zone between the red and white PAPI signals. Analysis 54 Aircraft Accident Report This analysis will focus on the flight crewmembers’ decisions and performance— as a crew and individually—to consider why none of the three flight crewmembers recognized that the airplane was below the glidepath during about the last 40 seconds of the approach and responded accordingly. 2.2 The Accident Approach Postaccident interviews with the pilots and examination of the CVR and FDR data indicated that the en route portion of the accident flight from MEM to TLH was routine and that the pilots engaged in normal duties and discussions as the airplane neared TLH. CVR evidence indicated that although the flight crew originally planned to land on runway 27, when they were about 10 minutes from touchdown, they decided to land on runway 9 instead. According to the Safety Board’s airplane performance study, when the airplane became established on the final approach course about 2 1/2 miles from the approach end of runway 9, the PAPI lights would have shown a low indication (one white and three red lights). Almost immediately thereafter, the PAPI lights would have shown a very low indication (four red lights), which would have been viewable from the cockpit for the remainder of the flight. As the airplane descended through 500 feet agl at 1,248 fpm, 152 knots, and with engines operating at about 1.17 EPR, the captain announced that the approach was “stable.” The Safety Board notes that, although the airplane’s airspeed was within the target range, the airplane did not meet FedEx’s criteria for a stabilized approach because its rate of descent was greater than FedEx’s recommended 1,000 fpm,106 the engines’ power settings were less than the expected 1.3 to 1.45 EPR, and its glidepath was low as indicated by the PAPI light guidance. According to FedEx procedures at the time of the accident, if a visual approach was not stabilized when the airplane descended through 500 feet agl, the pilots were to perform a go-around. The Safety Board concludes that the accident approach was not stabilized as the airplane descended through 500 feet agl and that the pilots should have detected this and performed a go-around. As the pilots continued the approach, the flight engineer read the before landing checklist query items, with the captain responding, in accordance with FedEx procedures.

[ANALYSIS] Research has shown that in situations like this, a pilot typically flies a lower-than-normal approach until the error starts to become apparent (usually about 2 to 3 miles from the runway), at which point the pilot takes corrective action. Indeed, on the night of the accident, the first officer did fly a concave approach, with a steeper-than-normal initial descent, which is characteristic of a black hole approach. However, in the case of the accident flight, the first officer did not modify the steepness of the approach path in time, and the airplane collided with trees and terrain. The Safety Board concludes that the approach to runway 9 at TLH (which was flown over unlighted terrain and in night visual conditions) resulted in black hole conditions, which likely contributed to the flight crew’s failure to properly perform the approach. However, the Safety Board also concludes that PAPI lights, such as those installed at runway 9 at TLH, are a recognized countermeasure for use in black hole conditions and should have been, but were not, effectively used to maintain an appropriate glidepath by the first officer (who was the flying pilot) or by the captain and flight engineer (who, under the principles of basic crew coordination, were in a position to receive this information and initiate a corrective response).108 107 FedEx records indicate that the captain and flight engineer had not received this training. In the captain’s case, he missed the black hole recurrent training module because he attended upgrade training instead. The flight engineer had not been with the company long enough to attend recurrent training. 108 The Safety Board also considered the possibility that the 0.4 percent upslope on the first third of runway 9 contributed to the flight crew’s low approach. The geometric appearance of upsloping runways can create the illusion that an airplane’s approach path is higher than desired, and a pilot might compensate by flying a lower-than-normal approach. However, in this case, the upslope was minor and only present for the first third of runway 9, and the remainder of the runway was a gradual downslope. Therefore, it is unlikely that the illusion created by the runway’s slight upslope contributed significantly to the pilots’ failure to maintain the correct glidepath. Analysis 56 Aircraft Accident Report 2.3 Fatigue Research109 on human alertness has shown that the early morning hours are often associated with degraded alertness and performance; the accident occurred about 0537 local time (0437 in the flight crew’s domicile time zone). However, because FedEx conducted most of its 727 domestic operations overnight, operating at these hours was likely a normal occurrence for the flight crew.

[ANALYSIS] Bouquet and others, “Color Discrimination Under Chronic Hypoxic Conditions (Simulated Climb “Everest-Comex 97”),” Percept Mot Skills Vol. 90, No. 1 (Feb 2000): 169-79. Analysis 65 Aircraft Accident Report Research shows that attention can be highly selective and that people may not respond to important objects that may be plainly visible.121 For example, a simulator study found that some pilots became so engrossed in performing a landing using a heads-up display that they failed to see an airplane blocking the end of the runway. Similarly, accident data confirm that an air crew can respond to a visual illusion of airport distance and fail to use accurate PAPI information that is directly visible. Fatigue and high workload are both likely to increase the possibility of missing relevant information. If the captain and flight engineer monitored the runway environment during the final approach and if the PAPI signal was visible without obstruction, inappropriate selective attention could help explain the failure of both crewmembers to respond to the PAPI information. The actions of the captain and first officer added to the complexity of the final approach and the need for careful monitoring by the flight engineer. For example, the captain decided, at the first officer’s suggestion, to change the landing runway, resulting in the performance of a nonprecision approach instead of a precision approach. The first officer was slow to correctly identify the airport, and the airplane was rather close to the airport before it was aligned on final approach. Thus, during the somewhat hurried final approach that ensued, the flight engineer might have served as a significant defense against an accident by carefully monitoring and crosschecking. The Safety Board notes that to accomplish the last item on the before landing checklist (landing lights), the flight engineer would need to look forward to check the landing light switches. According to FedEx procedures, the flight engineer should have remained facing forward, monitoring the approach for the remainder of the flight (in this case, 7 seconds). The flight engineer’s failure to adequately monitor the approach, either because of his workload, reliance on other crewmembers, inattention, or some combination of factors, removed the last defense against the accident. Further, research indicates that a lack of crew familiarity may have contributed to the flight crew’s failure to fly and monitor a stabilized approach.122 For example, in a study of major aviation accidents involving human performance issues, in which a large number of monitoring errors were observed, the Safety Board found that 73 percent of the accidents occurred on the first day that the captain and first officer had flown together; 44 percent occurred on the first flight leg. Simulator research supports this, showing that 121 See (a) A. Mack, “Inattentional Blindness,” Current Directions in Psychological Science Vol. 12, No. 5 (2003): 180-184. (b) F.W.

[CONCLUSIONS > FINDINGS] The first officer’s reported difficulty adapting to his schedule and frequently changing sleep cycles were conducive to the development of fatigue impairment that contributed to his degraded performance during the approach to Tallahassee Regional Airport; however, there were also other factors affecting the first officer’s performance (for example, his color vision deficiency). Conclusions 68 Aircraft Accident Report 11. It is possible that the flight engineer was impaired by fatigue at the time of the accident; however, it is also possible that the flight engineer’s poor monitoring of the late stages of the approach was the result of his workload during the somewhat rushed approach, the presumption that the two forward-facing flight crewmembers were adequately monitoring the approach, or some combination of factors. 12. The circumstances of this accident, in part, demonstrate the continuing need for fatigue management efforts similar to those being developed by the Department of Transportation Operator Fatigue Management Program in the aviation industry. 13. The first officer suffered from a severe color vision deficiency that made it difficult for him to correctly identify the color of the precision approach path indicator signal during the below-glidepath, nighttime, visual approach to runway 9 at Tallahassee Regional Airport. 14. Existing aviation medical certification standards for color vision and use of related screening tests may not ensure detection of color vision deficiencies that can be detrimental to safety; it is possible that in some emergency situations, the speed of color recognition may assume an importance that is not currently reflected in the standards. 15. One or more of the color vision screening tests currently approved for use in the aviation industry (for example, the Farnsworth Lantern screening test) are not adequate; these tests should be identified and their use discontinued. 16. The circumstances of this accident support the recent increase in emphasis on crew monitoring reflected in recent initiatives by the Federal Aviation Administration and aviation industry. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of the accident was the captain’s and first officer’s failure to establish and maintain a proper glidepath during the night visual approach to landing. Contributing to the accident was a combination of the captain’s and first officer’s fatigue, the captain’s and first officer’s failure to adhere to company flight procedures, the captain’s and flight engineer’s failure to monitor the approach, and the first officer’s color vision deficiency. 69 Aircraft Accident Report

[ANALYSIS] The airspeed increased about 5 kn as a result of an increase in the headwind component and the crew stated that the aiccraft rolled to the right. When the captain corrected the lateral motion, the aircraft ended up slightly high. Thus, the Board concludes that the encounter with the gust added to tne lift produced by the rotation of the aircraft and caused a prolonged float. The FDR data indicates that the aircraft floated about 4 seconds after the airspeed etabilized. The performance analyais shows that the deceleration during this period was normal for a 30° flap, idle thrust configuration. A review of the major events influencing the approach and landing ie neceasary to bring matters into perspective. The approach to the threshold was flown within normal variations of speed and altitude control and the aircraft was stabilized in the landing configuration. When the aircraft was over the threshold, the power levers were retarded according to procedures to érrive at the touchdown point with an idle power setting. FDR data and witness statements indicate that the aircraft's wheels were about 10 feet atove the runway at the 1,000-foot touchdown point, Thus, touchdown appeared to be imminent and there was no reacon for the captain to suspect that a go-around might be necessary. In fact, since the captain testified that he was "programmed to land,” his thought processes were probably oriented toward the next phase of the flight which was control of the atrcraft on the ground and the neceseary control inputs such as the power reversing process, braking, and steering of the aircraft. Thus, the captain's perceptions of the approach and iominent touchdown reinforced his expectation of another normal landing. ee ee eee ~ 29 - The next major even’. that occurred in this accident sequence was the encounter with the turbulence which caused a lateral upset of the a*trcraft of zufficient magnitude to cause either the captain or the first officer to utter an exclamation of surprise. The effect of the turbulence was sutstantiated by the FDR and the aircraft's reaction to it was observed by a qualified witness. The captain was now faced with the predicaxent of :eing unable to land the aircraft before taking the necessary action to correct tire lateral upset. The captain was also confronted with a more critical situation, The atrcraft was etill about 10 feet above the runway and well beyond the normal touchdown point. Thus, he was faced with en immediate decision to land the aircraft or to initiate a go-around. We probably had less than 5 seconds to evaluate the situation and take action as the aircraft was fast approaching the point along the runway froma which a go-around would have been a dangerous, if not an impossibie saneuver but from where the aircraft still could have beer landed and stopped successfully. These decisions which faced the captain represent the third mejor event iu this accident sequence. His decision to positively put the rircraft on the runway cave shortly after the first officer issued his "srill high, Are" warning. The pilot pushed over the nose of the aircraft and forced it onto the runway.

[ANALYSIS] The "bangs" recorded on the CVR tape from Flight 625 were compared by spectrum analysis with CVR tapes with known compressor stall noises from other B-727 aircraft. This analysis gave no ;ositive evidence of coapressor stalls on Flight 625. Additionally, investigative findings indicated that even had a few low magnitude compressor stalls occurred at the time the "bangs" were heard, the effect on engine performance would have been negligible. The fligttcrew was properly certificated and each crewmember had recejved the training and off-duty time prescribec by applicable regulations, There was no evide ‘ce of medical or physiological problems that might have affected their performances. The airport was properly certificated in accordance with 14 CFR 139 and there were no exemptions in effect on the day of the accident. 2.2 Approach nd Landing Witness observations, crev statements, FDR information, and CVR information all indicate that the aircraft approached the runway in a normal profile which would result in a touchdown 1,000 feet or slightly more beycnd the threshold. Instead of touching down, however, the aircraft floated 5 to 10 feet above the runway's surface. The FDR data indicate that the aircraft floated between 7 and 8 seconds, during which time it would have traveled about 1,500 feet. This correlated with witness observations which placed the touchdown between 2,500 and 2,900 feet beyond the threshold. After touchdown, the captain, who was concerned that he would not be able to stop the aircraft on the remaininy runway, decided to execute a go-around maneuver. He announced his irtention to go around about 3 seconds after touchdown. After the accident, the captain stated that he moved the thrust levers to a vertical position and hesitated in order to allow the engines to attain a stabilized thrust, about 1.4 EPR, before going to takec “£ power. This procedure is prescribed in the American Airlines operating manual for go-around from the runway. When the captain thought that the engines vere not accelerating at an expected rate, he roved the thrust levers tv the forward stop. Then, when it appeared to him that the engines were again not accelerating at a rate which would result in a successful go-around, he brought the thrust levers back tou idle and attempted to slow the aircraft by using maximum wheel braking. He did not at this time, employ any other braking devices. The activation of the takeoff warning horn disclosed that the captain moved the thrust levers to the vertical position at the same time that he announced his intention to go-around--within 3 seconds of the time of touchdown.

[ANALYSIS] The Safety Board believes that the existing regulations for the certification of a given air carrier operation are adequate. The regulations do not specifically consider factors particular to the airport, At the Harry S Truman Airport for example, there are know to be unique wind conditions which can cause an adverse deviation in an aircraft's performance. There are also visual factors related to runway dimensions and surrounding terrain. However, the regu)ations do include ~-~ 34 - a provision whereby any carrier, pilot group or individual pilot can halt operation into any airport and report to the proper authority any known specific hazard condition, The FAA witnesses testified that they were unaware of any such reports defore this accident. Furthermore, in evaluating the adequacy of the airport for jet aircraft operetions, one must review the accidents which have occurred. Since the beginning of commercial jet operations at Harry S Truman Airport in 1965, there had been two accidents, before this accident, involving air carrier jet aircraft, one of which included fatalities. These accidents involved a DC-9 in 1969 and a B-727 in 1970. The probable cause of the first accident was determined to be a loss of effective braking action caused by dynamic hydroplaning. There were several improvements to the airrort following that accident, including the grooving of the runway. These improvements were designed to help prevent hydroplaning accidents. The probable cause of the second accident, in which two persons died, was determined to be the captain's use of improper techniques in recovering fron a high bounce after a poorly executed approach and touchdown. The airport handles wore than 7,000 jet aircraft operations annually. Based on this evidence the safety Board concludes that the airport, although less than ideal, is safe with regard to B-727~100 operations, provided that these operations are conducted within escribed procedures. 2.4 Crash/Fire/Rescue A key factor in evaluating crash/fire/rescue response to an accident is how quickly was such a response required in order to save lives. Testimony by witnesses and surviving passengers indicate that the fires began when the right wing struck the embankment at the end of the runway. Even as the aircraft slid to a halt, the smoke had become dense in and around the fuselage. The flight attendant on the gatiley jumpseat stated that she saw the first-class cabin being torn apart in the impact sequence. Tie flight attendant in the rear of the aircraft indicated that when the aircraft came to a halt, the tail section in which she was sitting had broken off. Both flight attendants stated that as soon as the impac’. sequence stopped, and before they could leave their seats, they saw fire within 4 feet away. They stated that the flames were moving rapidiy toward then. In addition, they saw dense black smoke imacdiately, which made breathing difficult.

[ANALYSIS] This distance increases to 2,387 feet {{ff the pilot hesitates with the thrust levers at the vertical position. An inherent danger in the go-around maneuver is that the pilot will rotate the aircraft to the takeoff artitude before sufficient thrust has developed to counter deceleration. This procedure is likely to increase the distance required to ltft off even more than that determined by the analysis. ~ 26 ~- The Safety Board, thus, concludes that a successful go-around could not have been executed when the captain attempted to do so. Powever, the analysis of the aircraft's braking performance irdicated that, using maximum braking ana spoilers, the aircraft could have been stopped in less distance than required for go-around. In fact, after this tanding, the captain should have been able to scop the aircraft on the runwey and certainly within the confines of the runway overrun. Therefore, to complete the analysis of this accident, there are several areas which must be discussed. These are the factors and circumstances leading to the long touchdown, the decision making process of the captain before and after the landing, the factors which could have influenced this decision making process, and the adequacy of the margins of safety in the FAA's airport and aircraft certification criteria. In order to evaluate the factors and circumstances which led to the long touchdown, the extent and comprehenstveness of the company's operational guidelines nust be examined. Additionally, the extent to which the approach conformed to the specific training and operating inatructions given to American Airlines pilote must be reviewed. Company guidelines were set forth in a 1971 memorandum issued to all pilots who were flying American's Caribbean routes. The memorandum contained company policy concerning flap usage, aiming point, touchdow. point, and go-arounds. It pointed out the possibilities of encountering downdrafts on the approach, and emphasized the necessity of being in the "“slet", the importance of the 1,000-ft aiming point, and the possible existence of a wind shear which could produce a float if the aircraft is landed long beyond the 1,000-ft point. The memorandum also pointed eut the necessity of executing a go-around if the approach is not in the slot, if the landing will be “appreciably” beyond the 1,000-ft point, or if a bounce occurs on initial touchdown. The memorandum stated that "the use of 40° of flap is the standard practice;" however, there was an oytion to use either 30° or 40° flaps “with strong, gusty winds." It recommended the use of 30° flap with a wind component of 20 kn or more. A 1972 Operations Bulletin FM2 C-7, with one change, reiterated the guidelines on flap usage contained in the 1971 memorandum. In restating the policy regarding the option to use 30° or 40° flaps, the winds vere described as "with strong or gusty winds.” The word “or” is underlined in the bulletin.

[ANALYSIS] At the very least, a much lower velocity impact would have occurred. 6/ Davis, “Human Errors and Transport Accidents," Ergonomics, 2.24 (1958) Department of Medicine, University of Cambridge. eee enn paneer nie ~ 32 - The Board believes that intensive training is the most effective means to combat the effects of this emergency mechanism. Had the captain been exposed during training to critical go-around situations and to the maximum performance stopping capabilities of the aircraft by means of flight sfmulation and lectures, he may have reacted appropriately in this situation. In summary, it ie evident that the captain could have extracted himself twice from a dangerous situation and could heve avoided this accident. His first opportunity was during the turbulence encounter just after passing the 1,000-foot touchdown area; he should have followed company procedures and should have initiated a go-around as scon as he regained control of his aircraft. His second opportunity to avoid the accident came when he decided to land the aircraft; he should have applied maximum performance stopping procedures to bring the aircraft to a stop within the remaining runway length. Furthermore, when an accident was inevitable, the captain had an opportunity to lessen the damage to the aircraft and to diminish the impact velocity; at this time, he should have applied maximum performance stopping procedures. Thus, while the Safety Board beliaves that the causal area of this accident involves the captain's actions before and after the touchdown, his lack of substantive information about the aircraft's stop or go~around performance capabiiities seriously affected hie ability to make a proper decisio- in this situation. The Board is aware that American Airlines’ training procedures have been revised to include these performance factors. An Operations Bulletin (FH2 C-13) was issued on August 16, 1976, and placed in the Plight Manual. The bulletin states that any decision to go-around at St. Thomas should be made and initiated no later than the 1,000-foot touchdown markers. This bulletin ccntinues with the following company policy: "Go-around shall not be attempted after the aircraft has touched down on the runway, and the landing should be continued to a stop -- recognizing the full stopping capabilities of the 727 with spoilers, main and nose gear brakes." Thus, American Airlines recognizes that in their operations at St. Thomas, pilot knowledge of the go-around and stopping capabilities of the Boeing 727 are vital to a safe operation, The Board is concerned, however, that there may be other airports exhihitt-¢ similar operational and environmental conditions which are sezved by turbine-powered aircraft and by other air carriers. Therefore, to prevent a similar accident, these areas of concern have been addressed in a recommendation letter to the Federal Aviation Administration. 2.3 Airport Certification Finally, the Safety Board considered the adequacy of FAA aircraft and airport certification criteria for the Harry S Truman Airport. The requirements of 14 CFR 139 are confined exclusively to the airport proper. --- Footnotes: [6/ Davis, “Human Errors and Transport Accidents," Ergonomics, 2.24 (1958) Department of Medicine, University of Cambridge.]

[CONCLUSIONS > FINDINGS] The accident sequence took place near the edge of a thunderstorm. 51 11. The pilot in command decided to turn right immediately after takeoff to avoid the nearby thunderstorm and heavy precipitation that would have been encountered on a straightout departure. 12. The airplane was 96 pounds over the maximum gross takeoff weight at takeoff, and 84 pounds over the maximum gross takeoff weight at the time of the impact. 13. Although horizontal in-flight visibility at the time of the stall was most likely substantially degraded due to precipitation, eliminating a visible horizon, the pilot in command could have maintained visual ground reference by looking out the side window. However, this could have been disorienting to the pilot. 14. The airplane experienced strong crosswinds, moderate turbulence and gusty winds during its takeoff and attempted climb, and the pilot in command was aware of these adverse wind conditions prior to executing the takeoff. 15. The right turn into a tailwind may have caused the pilot in command to misjudge the margin of safety above the airplane’s stall speed. In addition, the pilot may have increased the airplane’s pitch angle to compensate for the perceived decreased climb rate, especially if the pilot misperceived the apparent ground speed for airspeed, or if the pilot became disoriented. 16. The high density altitude and possibly the pilot in command’s limited experience with this type of takeoff contributed to the loss of airspeed that led to the stall. 17. The pilot in command failed to ensure that the airplane maintained sufficient airspeed during the initial climb and subsequent downwind turn to ensure an adequate margin above the airplane’s stall speed, resulting in a stall and collision with the terrain. 52 18. The pilot in command inappropriately decided to take off under conditions that were too challenging for the pilot trainee and, apparently, even for him to handle safely. 19. The pilot in command suffered from fatigue on the day before the accident. 20. Information on fatigue and its effects, and methods to counteract it, might have assisted the pilot in command to recognize his own fatigue on the first day of the flight, and possibly enhanced the safety of the trip. 21. The airplane occupants’ participation in media events the night before and the morning of the accident flight resulted in a later-than-planned takeoff from Cheyenne under deteriorating weather conditions. 22. The presence of media at the Cheyenne Airport and media interviews scheduled for the next two overnight stops probably also added pressure to attempt the takeoff and maintain the itinerary. 23.

[ANALYSIS] Higher density altitudes result in a reduction of aerodynamic (wing and propeller) and powerplant performance during takeoff and initial climb, and in a longer takeoff run and slower rate of climb. This reduced rate of climb might well prompt a person who was inexperienced with high density altitude takeoffs to raise the nose of the airplane in an attempt to increase the rate of climb, thereby further decreasing the airspeed. Therefore, the Safety Board concludes that the high density altitude and possibly the pilot in command’s limited experience with this type of takeoff contributed to the loss of airspeed that led to the stall. The Safety Board has been unable to determine which of the above factors, or a combination of factors, resulted in the reduction in the climb speed to below the stall speed. However, the Safety Board concludes that the pilot in command failed to ensure that the airplane maintained sufficient airspeed during the initial climb and subsequent downwind turn to ensure an adequate margin above the airplane’s stall speed, resulting in a stall and collision with the terrain. The Safety Board notes that the pilot in command had limited experience operating out of high density altitude airports, such as Cheyenne, and that this should have prompted him to be cautious. He had expressed concern about the predicted storm that was to move in from the west, and he had wanted to leave early enough to avoid the storm. Further, just prior to departure, the pilot knew the wind conditions reported by a pilot who had just departed. Accordingly, the Safety Board concludes that the pilot in command inappropriately decided to take off under conditions that were too challenging for the pilot trainee and, apparently, even for him to handle safely. Therefore, the Safety Board attempted to analyze the human performance factors that might have caused the pilot in command to depart under those conditions. These factors include the possible effects of fatigue, the emphasis placed on media events, and the desire to adhere to the programmed itinerary. 41 2.3 Human Performance Aspects 2.3.1 Fatigue The pilot in command’s sleeping schedule in the days before the accident flight may have led to fatigue. He received 6 ½, 6 ¾, and 5 ½ hours of sleep, respectively, in the 3 days prior to the start of the trip on April 10, compared to the 8 ½ to 9 hours of sleep that he typically received per night on weekends.41 On April 10, he awoke at 0330, earlier than his normal wake-up time. By midafternoon on April 10, during the fueling stop at Rock Springs, he told a witness of being tired. After arriving at Cheyenne, he telephoned his wife and said that he “was really tired.” There is evidence that people tend to underestimate their level of tiredness,42 so that when the pilot reported being “really tired” it probably reflected a high level of fatigue.

[CONCLUSIONS] The pilot in command, pilot trainee, and rear seat passenger (the pilot trainee’s father) were fatally injured. Instrument meteorological conditions existed at the time, and a visual flight rules flight plan had been filed. The flight, which was a continuation of a transcontinental flight “record” attempt by the youngest “pilot” to date (the pilot trainee), was operated under the provisions of 14 Code of Federal Regulations Part 91. The National Transportation Safety Board determines that the probable cause of this accident was the pilot in command’s improper decision to take off into deteriorating weather conditions (including turbulence, gusty winds, and an advancing thunderstorm and associated precipitation) when the airplane was overweight and when the density altitude was higher than he was accustomed to, resulting in a stall caused by failure to maintain airspeed. Contributing to the pilot in command’s decision to take off was a desire to adhere to an overly ambitious itinerary, in part, because of media commitments. The safety issues discussed in the report include fatigue, the effects of media attention and itinerary pressure, and aeronautical decision making. A recommendation concerning the circumstances of this accident and the importance of aeronautical decision making was made to the Aircraft Owners and Pilots Association, the Experimental Aircraft Association, and the National Association of Flight Instructors. Recommendations concerning aeronautical decision making and the hazards of fatigue and were made to the Federal Aviation Administration. NATIONAL TRANSPORTATION SAFETY BOARD WASHINGTON, D.C. 20594 AIRCRAFT ACCIDENT REPORT IN-FLIGHT LOSS OF CONTROL AND SUBSEQUENT COLLISION WITH TERRAIN CESSNA 177B, N35207 CHEYENNE, WYOMING APRIL 11, 1996 1. FACTUAL INFORMATION 1.1 History of Flight On April 11, 1996, about 0824 mountain daylight time (MDT),1 a privately owned Cessna 177B, registration N35207, collided with terrain after a loss of control following takeoff from runway 30 at the Cheyenne Airport, Cheyenne, Wyoming. The pilot in command, pilot trainee,2 and rear seat passenger (the pilot trainee’s father) were fatally injured. Instrument meteorological conditions existed at the time, and a VFR3 flight plan had been filed.

[CONCLUSIONS > FINDINGS] The airplane occupants’ participation in media events the night before and the morning of the accident flight resulted in a later-than-planned takeoff from Cheyenne under deteriorating weather conditions. 22. The presence of media at the Cheyenne Airport and media interviews scheduled for the next two overnight stops probably also added pressure to attempt the takeoff and maintain the itinerary. 23. The itinerary was overly ambitious, and a desire to adhere to it may have contributed to the pilot in command’s decision to take off under the questionable conditions at Cheyenne. 53 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the pilot in command’s improper decision to take off into deteriorating weather conditions (including turbulence, gusty winds, and an advancing thunderstorm and associated precipitation) when the airplane was overweight and when the density altitude was higher than he was accustomed to, resulting in a stall caused by failure to maintain airspeed. Contributing to the pilot in command’s decision to take off was a desire to adhere to an overly ambitious itinerary, in part, because of media commitments. 54 4. RECOMMENDATIONS As a result of the investigation of this accident, the National Transportation Safety Board makes the following recommendations: --to the Aircraft Owners and Pilots Association, the Experimental Aircraft Association, and the National Association of Flight Instructors: Disseminate information about the circumstances of this accident and continue to emphasize to your members the importance of aeronautical decision making. (A-97-19) --to the Federal Aviation Administration: Expand the development and increase the dissemination of educational materials on the hazards of fatigue to the general aviation piloting community. (A-97-20) Incorporate the lessons learned from this accident into educational materials on aeronautical decision making. (A-97-21) BY THE NATIONAL TRANSPORTATION SAFETY BOARD James E. Hall Chairman Robert T. Francis II Vice Chairman John Hammerschmidt Member John J. Goglia Member George W. Black, Jr. Member March 11, 1997 55 5. APPENDIXES APPENDIX A INVESTIGATION AND HEARING 1. Investigation The National Transportation Safety Board’s Northwest Region was notified of the accident by the FAA’s Northwest Mountain Region Operations Center at 0758 Pacific daylight time on April 11, 1996. The Safety Board’s investigator in charge departed for the site on the first available commercial flight. Safety Board personnel from the South Central Field Office, Denver, Colorado, were the first investigators to reach the site, arriving shortly after noon. Investigation groups were formed for human performance and meteorology. Parties to the investigation were the Federal Aviation Administration, Cessna Airplane Company, and Textron Lycoming. 2. Public Hearing There was no public hearing held or formal depositions taken for this accident. 56 APPENDIX B TRANSCRIPT OF AFSS WEATHER BRIEFING Q Memorandum U.S.

[ANALYSIS] It is known that the precipitation at the time of the accident was at least partly, if not totally, frozen. Therefore, to the extent that the precipitation at the time of the accident sequence was frozen, the precipitation rate, and therefore the loss of lift, would be less. 37 witness statements indicate that the turn to the east was gradual; this is consistent with a bank angle of about 20 degrees. With the flaps at 10 degrees, a 20-degree bank angle would increase the stall speed about 3 mph, from about 59 mph for steady level flight to about 62 mph. Using the total weights of the clothed occupants, a full load of fuel, personal baggage and miscellaneous items (food and fluids) plus the airplane’s basic empty weight, the Safety Board concludes that the airplane was 96 pounds over the maximum gross takeoff weight at takeoff, and 84 pounds over the maximum gross takeoff weight at the time of the impact. The additional weight would have increased the airplane’s stall speed about 2 percent. The combined effect of the rain, the excess weight and the bank angle of the airplane could have increased the airplane’s stall speed to about 64 mph. The airplane’s best rate of climb speed at sea level is 87 mph (with a climb rate of 685 feet per minute). However, the high density altitude at the airport and the overweight condition of the airplane would have affected the airplane’s climb performance. The Cheyenne Airport has a field elevation of 6,156 feet msl. Taking into account the temperature, the density altitude at the time of the takeoff and accident was 6,670 feet msl. According to airplane performance data from Cessna, the high density altitude and the airplane’s overweight condition would have decreased the airplane’s best rate-of-climb speed to 81 mph; with a climb rate of 387 feet per minute. Thus, the combination of the effects of the rain, the overweight condition, and the gradual bank angle of the airplane would have increased the airplane’s stall speed from about 59 mph to about 64 mph, and, along with the high density altitude, decreased its best rate-of-climb speed from 84 mph to about 81 mph. However, the airplane should have been able to climb and turn safely. Thus, the Safety Board analyzed the possible reasons why this did not occur. These include a possible reduction in engine power from carburetor icing or an over-rich fuel/air mixture, and the effects of fluctuating winds, poor visibility and the lack of sufficient experience in takeoffs from high density altitudes on the pilot in command’s ability to operate the airplane. As noted in section 1.6.1, it is necessary to lean the fuel mixture at higher altitudes to allow maximum engine performance; an over-rich mixture can 38 result in an appreciable loss of power and reduced climb performance capability. The mixture control knob was found in the full rich (forward) position at the accident site, thus suggesting that it was in that position prior to impact, although it is possible that the knob was out and impact forces moved the knob forward without bending the rod.

[ANALYSIS] However, according to his wife, he was pleased by the media attention. The media interest apparently began from an article in a local paper. Media interviews began several days before the trip, and included numerous telephone calls at home and early morning live television interviews on national news programs. Media representatives were present at nearly every stop on the trip, and the time spent participating in media events interfered with other activities that could have affected flight safety.48 For example, the media events at Cheyenne the night before the accident resulted in a delayed arrival at the hotel, thus delaying the opportunity for the participants to obtain much needed rest. Also, on the morning of the accident, the pilot in command’s preflight inspection and preparation of the airplane was interrupted by a television reporter’s 48In addition, the father videorecorded parts of the first day’s flight from the back seat of the airplane for use by a national news service, providing a regular reminder of the news value of the flight. 45 interview, possibly degrading the thoroughness of the preflight preparations.49 The airplane occupants also participated in two other media interviews, further delaying the departure. Further, slips of paper found in the shirt pocket of the pilot trainee’s father indicated that they had media interviews scheduled for that evening in Ft. Wayne, Indiana, and the following evening in Massachusetts. On the morning of the accident, the weather was changing. There was a cold front approaching Cheyenne, but the weather to the east was good. If the airplane had departed ahead of the storm, it would have encountered good weather along its intended route. If the airplane had waited until after the storm passed, its intended route would have been blocked by the storm. The pilot had made it clear the night before that he wanted to depart ahead of the cold front, but their departure was delayed by a later-than-planned checkout from the hotel and by the media interviews. By the time the airplane was ready for takeoff, the only way to avoid being held up by the storm, and therefore to maintain the tightly scheduled itinerary, was to fly for several minutes in the deteriorating and potentially unsafe weather conditions associated with the storm. Self-induced pressures from media attention can degrade decision making, increasing the perceived importance of maintaining the schedule compared to other factors. It would have been a normal human response for the pilot in command to have been affected by the media attention. The Safety Board concludes that the airplane occupants’ participation in media events the night before and the morning of the accident flight resulted in a later-than-planned takeoff from Cheyenne under deteriorating weather conditions.

[ANALYSIS] Thus, the actual wind change acting on the airplane could have been as much as 6 knots per second over a 10- to 12-second period. The wind velocities necessary to produce such a change are not unusual in microburst conditions. However, without more definitive data regarding the timing of the captain's actions to advance power, to retract the landing gear, and to raise the flaps, the precise wind environment of the thunderstorm cannot be reconstructed. Nevertheless, the Safety Board concludes that the wind shear was severe and that the captain's actions to initiate the missed approach probably prevented a more catastrophic accident. It is likely that the captain's action to start a missed approach when the airplane entered the rain and hail shaft prevented the airplane from hitting the ground well short of the runway. As the airplane exited the significant rain and hail associated with the thunderstorm, the captain apparently became concerned that the missed approach would not be successful, and he decided that the best course of action would be to put the ~34- airplane on the ground. According to the captain, his decision was based upon his perception that the airspeed had decreased to 119 knots and that the airplane was still descending. Although the UFDR showed that the airplane was maintaining an altitude about 120 feet above the ground, the increase in visual cues as the airplane exited the rain may have heightened a perception that the airplane was too low and caused him to believe that descent to the ground could not be avoided even with maximum power. Given the captain's perception of possibly inevitable contact with the ground, the lowering of the landing gear was appropriate. However, the Safety Board believes that the captain's action to abandon the missed approach and reduce power was improper. There were not sufficient data available regarding the timing of the captain's actions to extend the landing gear and reduce power to permit an accurate analysis of the relative effect of the wind and pilot actions. Additionally, the insufficient data precluded an analysis of the airplane's performance and the possibility that descent below 120 feet and even contact with the ground could have been avoided had the missed approach been continued. However, the UFDR data showed that the airspeed began to increase as the descent to the runway was initiated. Given the captain's statement that he had reduced power, the airspeed rise supports the hypothesis that the airplane had penetrated the severe wind shear and had achieved climb capability when it struck the runway. A well-trained and alert captain should have been aware that the airplane at maximum power would regain a positive performance capability after penetrating a microburst. Moreover, he should have anticipated that the airplane would begin to respond to maximum power after it exited the rain and hail shaft. Even though his restricted forward visibility may have precluded his awareness that over 8,000 feet of runway remained, his knowledge that he was over the runway should have mitigated his concern about possible ground contact after the landing gear were down during the continued missed approach.

[ANALYSIS] Based on the factors cited above, the Safety Board concludes that the captain's decision to continue the instrument approach to the point where the missed approach was started was imprudent, showed poor judgment, and subsequently resulted in a severe wind shear encounter which led to the accident. The final operational decision of concern was the decision to land immediately, which the captain made when he saw the runway. Shortly after Flight 183 passed the middle marker at 1656:16, the airplane may have had the capability to continue to maintain a constant, level altitude. However, the captain apparently did not consider further efforts to continue the missed approach once he saw the runway and, in fact, said that, "the only thing that I could think of was the landing, contact to the ground was imminent." Consequently, the captain's decision, made sometime between 1656:16 and 1656:21, was to exercise what he believed to be his only option, which was to land on the runway. The captain lowered the pitch attitude of the airplane, reduced thrust, and lowered the landing gear. The sound similar to stickshaker 3 seconds before impact was likely caused by the flare of the airplane to halt the rate of descent and to cushion the landing on the runway. It was not caused by an attempt to fly the airplane at stickshaker airspeed. The airplane performance analysis established that although the indicated airspeed was decreasing steadily, the airplane was not descending or "being pushed down" as the captain recalled. Rather, the captain's perception of what was happening to the airplane was likely influenced by his perception of the need to apply an abnormal force to the control wheel to raise the nose of the airplane, by the intimidating sounds of hail and heavy rain on the cockpit, and, most significantly, by the observed loss of 24 knots of airspeed in 10 to 12 seconds. As a result, when the captain saw the runway ahead of him, he immediately committed the airplane to the runway. The flighterew's recollection of the accident parallels closely findings included in an analysis of a number of wind shear-related accidents and incidents that was made by the Boeing Commercial Airplane Company. 16/ In particular, Boeing's analysis found that pilots experienced difficulty in maintaining proper pitch attitudes because of "perceived" heavy elevator control forces. The analysis stated, "The pull stick forces required, after a normal rotation, may be perceived as an abnormal response and a concern to a pilot expecting to be flying in trim during climb." The analysis also noted that, "Even the decreasing pitch attitude in response to the loss of airspeed may be perceived as normal in light of the pilot's airspeed control training emphasis." The analysis concluded that since pilot reaction to wind shear encounters must be made in seconds when at a low altitude, it was difficult for the flying pilot alone to scan, interpret, and react to all the cockpit information in a timely fashion. Therefore, specific wind shear. training, emphasizing coordinated flighterew actions and reactions, and proper flight path management were critical to minimize the hazards of the wind shear encounter.

[ANALYSIS] Moreover, he should have anticipated that the airplane would begin to respond to maximum power after it exited the rain and hail shaft. Even though his restricted forward visibility may have precluded his awareness that over 8,000 feet of runway remained, his knowledge that he was over the runway should have mitigated his concern about possible ground contact after the landing gear were down during the continued missed approach. Therefore, rather than reduce power and commit to a landing before the landing gear were fully down, the Safety Board believes that the proper decision would have been to continue the missed approach even after the landing gear were lowered and even if a "touch and go" on the runway proved necessary to prevent further loss of airspeed. 2.4 Operational Factors The Safety Board believes that the major safety lesson from this accident involves the flighterew's decisions made at the beginning of the approach which led to entering the microburst with its wind shear rather than in an evaluation of the crew's performance after the wind shear encounter or the ability of the airplane to penetrate the microburst successfully. The key issues in the accident sequence thus were the operational factors, which, combined with meteorological and airplane performance factors, influenced the captain's decision to initiate and continue the instrument approach to runway 21R. To address these issues, the Safety Board analyzed the policies and guidelines of USAir which addressed adverse weather, the role of air traffic control, and the actions of the flightcrew during the instrument approach. USAir Policy and Procedures.--The USAir DC9 Pilot's Handbook guidelines for thunderstorm and wind shear encounters are supported by topical material in the company flighterew publications (Flight Crew View and Flight Crew Quarterly). The company publications expand on handbook policy and procedures and provide substantial information on weather phenomena. Additionally, the thunderstorm and wind shear information parallel data are contained in FAA Advisory Circular AC 00-50A, "Low Level Wind Shear." The information and guidance provided USAir pilots in company manuals and handbooks were accurate and emphasized that wind shear and gust fronts associated with thunderstorms can precede the actual storm by as much as 15 miles. Additionally, the Pilot's Handbook provides specific guidance for configuration of the airplane if a wind shear may be encountered and for flying the airplane at stickshaker speeds. USAir does not have a specific policy which governs the avoidance of thunderstorms in the terminal area. The Pilot's Handbook, under the weather radar section, states’ that storm cells should be avoided by 5 miles when the outside air temperature is above freezing. However, the 5-mile guideline is not absolute. The Director, Flight Training and Standards stated that the captain's judgment in a landing situation takes precedence over the 5-mile guideline in the handbook. Additionally, while the handbook warned of thunderstorm-related hazards within 15 miles of the actual storm, the information was regarded as advisory in nature and not a definite or suggested limit.

[ANALYSIS] As the airplane descended through about 800 feet m.s.1., the airspeed began to rise and the descent rate reduced. The airspeed rose from a stabilized level of 135 knots to about 143 knots in 4 seconds, after which it decreased to a low value of 119 knots within the next 12 seconds. The airplane remained leve] about 120 feet above the ground as the speed decayed. The Safety Board concludes that the increase in airspeed and the reduction in the descent rate as the airplane reached 800 feet m.s.1. was caused by the encounter with the outflowing winds from the center of the microburst. As the airplane continued to penetrate the divergent wind field, it encountered a sudden wind shift from the airplane's nose to its tail. The sudden reduction in the headwind component produced the airspeed loss. The longitudinal stability characteristics of an airplane encountering these conditions would cause the airplane to pitch nose down and accelerate to regain the ’ neutral trim airspeed. The pitch-down tendency must be countered by pilot force exerted to the control column and possibly pitch trim changes if the wind shear is to be penetrated without a significant loss of altitude. In fact, the procedures recommended for pilots upon a sudden inadvertent entry into a microburst-type wind shear during approach (below 500 feet) includes the immediate advancement of power to a maximum thrust level and the initiation of go-around procedures. The fact that Flight 183 remained at a nearly level altitude as the airspeed decayed showed that the captain did respond appropriately with the increasing pitch-up control forces necessary to prevent a continuing descent. Further, the power increase noted by the flight attendants and passengers and the airplane configuration when it struck the ground are compatible with the captain's action to initiate a missed approach. The Safety Board could not determine precisely when the go-around was initiated in the sequence of events as indicated by the UFDR data. However, it is evident that the airplane did not achieve a positive rate of climb even with maximum power. A DC9 under the existing conditions of weight, configuration, and density altitude should have been capable of a level flight inertial acceleration of about 3.9 knots per second with maximum power. Under a stable wind condition, this inertial acceleration would be reflected directly in an equivalent increase in airspeed. However, the airspeed actually decreased at about 2 knots per second, indicating that the longitudinal component of the wind along the airplane's flightpath was changing at a rate about 2 knots per second greater than the airplane's longitudinal acceleration capability. Thus, the actual wind change acting on the airplane could have been as much as 6 knots per second over a 10- to 12-second period. The wind velocities necessary to produce such a change are not unusual in microburst conditions.

[ANALYSIS] However, the 5-mile guideline is not absolute. The Director, Flight Training and Standards stated that the captain's judgment in a landing situation takes precedence over the 5-mile guideline in the handbook. Additionally, while the handbook warned of thunderstorm-related hazards within 15 miles of the actual storm, the information was regarded as advisory in nature and not a definite or suggested limit. Perhaps the most specific guidance provided flighterews was contained in a 1983 article on thunderstorms in the Flight Crew View, which stated: Do's and Don'ts of Thunderstorm Flying: Don't land or take off in the face of an approaching thunderstorm. A sudden gust front of low level turbulence could cause loss of control. Similar articles are consistent in stating that USAir does not want its flightcrews to take off, land, or fly near thunderstorms. However, the same strong guidance and admonitions are not contained in the DC9 Pilot's Handbook--the policy manual. Consequently, although there was no specifie prohibition by USAir policy governing a landing in the proximity of the thunderstorm, the intent of the company handbook and publications is evident. Flightcrews are expected to anticipate the hazards of thunderstorms, to evaluate the available information, and to make a decision based on the safety of the flight. The policies and procedures in the DC9 Pilot's Handbook were straightforward, including the procedures related to the rotation to stickshaker in an emergency. The USAir training program exposed pilots to the maneuver and explained the additional performance capability that may be available in an emergency if an airplane is rotated to stickshaker. The Safety Board believes that pilots should be made aware of the performance available by rotating to stickshaker as a last resort during a severe wind shear encounter. However, the maneuver must be performed in a flight simulator to give pilots the proper foundation to anticipate and perform the maneuver in an actual wind shear encounter. Air Traffic Control.—The quality and management of the air traffic services was related directly to the workloads of the controllers at the time of the accident. Controllers and supervisory personnel described the traffic and workload as moderate to heavy and "very" complex because of the number of aircraft and the developing weather conditions. Heavy workloads are likely to cause controllers to accelerate the pace of their transmissions and to increase the chance of transmission errors or omissions. Each of these circumstances arose at the radar and local control positions in the sequence of events which preceded the accident. At 1651:51, the East Arrival Radar controller cleared Flight 183 for the ILS approach to runway 21L and said "USAir 183 six miles from Scofi, maintain three 'til Nesbi, reduce and maintain 170 to the marker." However, at the conclusion of the transmission he said "... correction, that's the ILS twenty one right USAir 183." Aside from the fact that the flighterew did not hear the correction to the initial approach clearance, some information in the clearance changed when the controller revised the landing runway to 21R.

[ANALYSIS] However, the captain’s failure to stop or slow the airplane’s descent indicates that he was not aware of the airplane’s excessive descent rate and/or significantly misjudged its proximity to the ground. Although current FAA regulations permit pilots to descend below the MDA to 100 feet above the TDZE after they observe the approach lights,107 such a descent may not be advisable under all circumstances. During an approach at night, in reduced visibility, approach lights alone would not provide an adequate sight picture for the pilots to make an appropriate approach to the runway. With only an approach light or lights for approach path reference, a pilot could focus on those lights while flying into the ground. Pilots who had flown into IRK at night told investigators that, other than the airport’s lighting systems, there were minimal lights or ground references beneath the approach course that would have helped the pilots judge the airplane’s position relative to the runway or height above terrain.108 105 It is not clear what the captain was looking at when he reported seeing the ground. On the basis of IRK ASOS weather observations and other pilot reports, it is likely that the airplane was skimming the bottom of the clouds as it descended through the MDA. The approach path was over farmland and woods, and postaccident interviews with local residents and other pilots indicated that there were few, if any, light sources (except the airport lights) or other ground references that were visible at night in the area. 106 At the MAP, if the pilots do not see the approach lights or runway environment, company and published approach procedures require a missed approach. 107 For additional information, see section 2.4. 108 On the basis of the airplane’s altitude, its distance from the runway, the relative brightness of the approach lights, and the fact that the VASI glidepath does not intercept the MDA until closer to the airport, the VASI lights would have appeared solid red and therefore would not have provided the pilots with much usable information. Analysis 47 Aircraft Accident Report The Safety Board concludes that the pilots failed to follow established procedures to effectively monitor the airplane’s descent rate and height above terrain during the later stages of the approach and relied too much on minimal external visual cues. Although descent rate and altitude information were readily available through cockpit instruments, both pilots were largely preoccupied with looking for the approach lights. 2.3 Crew Performance/Professionalism Cockpit communications recorded by the CVR indicated that the pilots frequently engaged in conversation that lacked a professional tone during the accident flight. The Board considered whether these unprofessional communications (some of which were made during a critical phase of flight below 10,000 feet msl and therefore were not consistent with FAA and Corporate Airlines sterile cockpit procedures) were a factor in this accident. CVR evidence indicated that the pilots appeared to be comfortable with each other and enjoyed flying and joking together. To an extent, this working relationship might have been a benefit in the cockpit.

[ANALYSIS] According to the FAA, published nonprecision approach procedures ensure obstacle clearance throughout the descent to the MDA but do not ensure obstacle clearance below this altitude. Therefore, it is critical that pilots refrain from descending 111 Additionally, use of the constant-angle-of-descent approach technique would eliminate the need to descend to 100 feet above the touchdown zone before reaching the runway. In marginal weather conditions, the procedure permitting descent to 100 feet above the TDZE could have a potential for attracting pilots into an unsafe environment. 112 Evidence (interviews with other Corporate Airlines pilots, examination of Corporate Airlines’ operations, and review of FDR data from the accident pilots’ previous approach to IRK) indicates that the accident pilots were aware of the Federal regulations allowing them to descend below the MDA if they had the approach lights in sight and took advantage of it if necessary. Analysis 51 Aircraft Accident Report below the MDA unless at least one of several very specific approach lighting or runway environment items is “distinctly visible and identifiable” to them, to provide guidance to the runway. This accident occurred after the pilots continued their descent through the MDA and even through 100 feet above the TDZE (1,064 feet msl), without adequate visual reference to the approach lights or runway environment. At night with low ceilings, reduced visibility, and limited visual cues (ground lights, etc.), it would have been very difficult for the pilots to detect obstacles along the approach path through external visual cues, let alone visually assess the airplane’s descent rate, distance from the airport, and height above the ground. The pilots (whom CVR evidence indicated wanted to land at IRK at the end of a long day and who likely began to see that they were breaking out of the clouds around the MDA) were apparently motivated to descend below the MDA to acquire visual cues that would allow them to continue the approach. The accident airplane initially impacted trees about 996 feet msl, which was about 62 feet below the “100 feet above TDZE” altitude. The Safety Board concludes that current regulations permitting pilots to descend below the MDA into a region where obstacle clearance is not assured results in reduced margins of safety for nonprecision approaches, especially in conditions of low ceilings, reduced visibility, and/or at night. Further, these regulations can have the unintended effect of encouraging some pilots to descend below the MDA in an attempt to acquire visual cues that will permit them to continue the approach, as occurred in this case.

[CONCLUSIONS > FINDINGS] 3.1 Findings 1. The captain and first officer were properly certificated and qualified in accordance with, and had received the training and rest time prescribed by, Federal regulations and company requirements. The flight crewmembers possessed valid and current medical certificates appropriate for 14 Code of Federal Regulations Part 121 flight operations. 2. The accident airplane was properly certificated and maintained and was equipped and dispatched in accordance with applicable regulations and industry practices. There was no evidence of any preexisting powerplant, system, or structural failure. 3. The accident airplane’s cargo and its loading were not factors in the accident. 4. Although the weather did not cause this accident, the low ceiling and reduced visibility at Kirksville Regional Airport made the nonprecision approach more challenging. 5. Corporate Airlines’ policies, procedures, and training were consistent with industry standards. 6. Air traffic control was not a factor in the accident. 7. The emergency response did not adversely affect the survivability of this accident. 8. An enhanced ground proximity warning system (required by Federal regulation since March 29, 2005), would have provided the pilots with a “too low terrain” alert in sufficient time to avoid collision with the trees. 9. Although the accident airplane’s 1,200 feet per minute (fpm) rate of descent was consistent with company procedure, it varied from current Federal Aviation Administration guidance that recommends a descent rate of no more than 1,000 fpm below 1,000 feet above ground level. 10. The pilots failed to follow established procedures to effectively monitor the airplane’s descent rate and height above terrain during the later stages of the approach and relied too much on minimal external visual cues. Although descent rate and altitude information were readily available through cockpit instruments, both pilots were largely preoccupied with looking for the approach lights. Conclusions 57 Aircraft Accident Report 11. The pilots’ nonessential conversation below 10,000 feet mean sea level was contrary to established sterile cockpit regulations and reflected a demeanor and cockpit environment that fostered deviation from established standard procedures, crew resource management disciplines, division of duties, and professionalism, reducing the margin of safety well below acceptable limits during the accident approach and likely contributing to the pilots’ degraded performance. 12. Compliance with sterile cockpit rules may have resulted in an increased focus on standard procedures and professionalism during the accident flight. 13. The captain should have, but did not, arrest the airplane’s rapid descent when they reached the minimum descent altitude (MDA), and the first officer should have, but did not, challenge the captain’s descent below the MDA. 14. The use of a constant-angle-of-descent approach technique, with its resultant stabilized, moderate rate-of-descent flightpath, and obstacle clearance, would have better positioned the accident airplane for a successful approach and landing. 15. Current regulations permitting pilots to descend below the minimum descent altitude (MDA) into a region where obstacle clearance is not assured may result in reduced margins of safety for nonprecision approaches, especially in conditions of low ceilings, reduced visibility, and/or at night.

[CONCLUSIONS] Current regulations permitting pilots to descend below the minimum descent altitude (MDA) into a region where obstacle clearance is not assured may result in reduced margins of safety for nonprecision approaches, especially in conditions of low ceilings, reduced visibility, and/or at night. Further, these regulations can have the unintended effect of encouraging some pilots to descend below the MDA in an attempt to acquire visual cues that will permit them to continue the approach, as occurred in this case. 16. On the basis of the less than optimal overnight rest time available, the early reporting time for duty, the length of the duty day, the number of flight legs, the demanding conditions (nonprecision instrument approaches flown manually in conditions of low ceilings and reduced visibilities) encountered during the long duty day (and the two previous days), it is likely that fatigue contributed to the pilots’ degraded performance and decisionmaking. 17. Existing Federal Aviation Administration pilot duty regulations do not reflect recent research on pilot fatigue and sleep issues, increasing the possibility that pilots will fly in a fatigued condition. 18. Providing pilots with additional fatigue-related training, such as that being developed by the Department of Transportation Operator Fatigue Management Program, may increase their awareness and use of fatigue avoidance techniques and thus improve safety margins. 19. Capturing the maximum recorded data possible is necessary for a more effective reconstruction of the events that lead to accidents and the issuance of more timely safety recommendations to prevent similar accidents from recurring. Conclusions 58 Aircraft Accident Report 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of the accident was the pilots’ failure to follow established procedures and properly conduct a nonprecision instrument approach at night in instrument meteorological conditions, including their descent below the minimum descent altitude (MDA) before required visual cues were available (which continued unmoderated until the airplane struck the trees) and their failure to adhere to the established division of duties between the flying and nonflying (monitoring) pilot. Contributing to the accident were the pilots’ failure to make standard callouts and the current Federal Aviation Regulations that allow pilots to descend below the MDA into a region in which safe obstacle clearance is not assured based upon seeing only the airport approach lights. The pilots’ failure to establish and maintain a professional demeanor during the flight and their fatigue likely contributed to their degraded performance. 59 Aircraft Accident Report

[ANALYSIS] Both pilots’ attitudes and inattention during subsequent operations demonstrated a lack of regard/respect for their responsibilities and duties. Despite the pilots’ unprofessional verbal behavior throughout much of the flight, CVR, FDR, ATC, and radar information indicated that they were generally attentive to required flight-related duties until shortly before the accident. For example, the pilots were very attentive to the weather conditions (low ceilings and reduced visibility) at IRK, checking and rechecking the IRK ASOS observations as they approached the airport. Additionally, the captain provided a thorough landing briefing in which he stated the MDA and runway TDZE (1,320 and 964 feet, respectively) and reviewed the missed approach procedures. He also restated the MDA as he began the approach. However, despite their apparent awareness of proper approach procedures and altitudes, the pilots continued their 1,200 fpm descent below the MDA without an appropriate visual reference. The captain should have leveled off and focused on the flight instruments and the first officer should have instructed him to level off. It was apparent that they did not have the requisite visual cues needed to descend further at that time. Instead, about 2 seconds after the airplane descended through the MDA, still descending about 1,200 fpm, the captain asked the first officer, “what do you think?” About 4 seconds later, the first officer responded, “I can’t see [expletive].” Corporate Airlines’ procedures dictated that the first officer was to monitor the approach and report deviations from standard company approach procedures. During the accident approach, the first officer occasionally provided the captain with appropriate nonflying/monitoring pilot support (such as when he prompted the approach checklist, provided DME callouts, repeated the MDA altitude, and prompted landing flaps). However, there were several instances in which the first officer failed to provide company-required callouts or provided nonstandard callouts. For example, he was required to callout “100 feet above minimums” and “minimums,” as the airplane descended through those altitudes, but instead he made a single callout of “thirteen twenty” as the airplane approached the MDA. Most significantly, the first officer did not challenge the captain’s failure to level off at or reduce the airplane’s rate of descent around the MDA, despite the fact that the first officer did not see any ground references and was responsible for monitoring the progress of the approach.110 110 The Safety Board notes that careful adherence to standard procedures and division of responsibility in the cockpit can significantly help pilots limit the degrading effects of fatigue. The role of fatigue in this accident is discussed in section 2.5. Analysis 49 Aircraft Accident Report The first officer appeared to be engaged in appropriate nonflying/monitoring pilot activities during the later stages of the approach; he was looking for external visual references and stated, “in sight, continue” when he eventually observed the approach lights (about 11 seconds after the airplane descended through the MDA). He also prompted the captain regarding the landing flap configuration late in the approach.

[ANALYSIS] Further, some witnesses stated that they lost sight of the airplane after it flew away from them, and before it struck the ground. Since the afrplane struck the ground about 3,600 feet from the witnesses, this would indicate that horizontal visibility was less than 1 mile. None of the witnesses observed the landing gear or other features of the airplane distinesly. Some only saw lights but others definitely saw the airplane fuselage and wings. Based on the assessment of horizontal and vertical visibility at the airport, on the fact that the pilot began the missed approach procedure, and on the observation of the witnesses that they did not see the afrolane for more than a glimpse, the airplane must have been quite low. Also, based 01 a consensus of the witness observations, and on the weather eonditions, it appears ttiat Flight 104 probably passed over the runway, in the vicinity of the VOR, between 200 and 300 feet. The probability that the airplane was below the MDA for the approach wher observed by the witnesses could be accounted for by the possibility that the pilot saw the runway lights, descended below the MDA for the landing, and then elther lost sight of the runway of realized that he could no: land, because he was not aligned or was too far down the runway. Regardless of the reason for the low altitude when passing the terminal, the pilot should have been able to complete a successful missed approach and the airplane should have stained considerable altliude before It arrived at the point where It crashed. In order to analyze tha possible explanations for the loss of altitude, the Safety Board calculated the rate of descent required to descend from the altitudes observed by the witnesses, when the airplane passed over the alrport to the point of initlal ground contact. The straight-lire distance from the VOR (missed approach point) to initkal ground contact was sbout 3,250 feet. The afr:peed assumed for the calculations was 120 KIAS, which was about 150 miles par hour true airspeed, or 220 feet per second (fps). Using these distances and speeds, the elapsed time from abeam the VOR to ground contact would have been about 14.5 seconds. Since she airplane turned right from about 030° to about 060° during the missed approach, and essuming a standard rate turn (3° per second), it would have required 10 seconds to compiete the turn. Therefore, tha actual ground track would have been en are and the actu) distance from the VOR to ground contact would have been greater than 3,250 feet. Consequently, the elapsed time from Yor passage to impact would have been slightly over 15 seconds for the assumed 120 KIAS, Using the conservative time of 15 seconds, the rate of descent from 300 feet above the alrport to impact would have been 1,20) feet per minute (fpm), or 20 fps; from 200 feet it would have been 800 fpm or 13.3 fps; and from 100 feet it would have been 400 fpm, or 6.6 fos.

[ANALYSIS] Since it cannot be determined what the actual descent profile was, these are average rates of descent over the entire distance from the VOR to impact. 2.2 Possible Reasons for tire Descent The Safety Board was not able to Jetermine why the airplane lost altitude and crashed during the missed approach attempt. Several of the more likely possibilities were considerad, but none of tham could be confirmed based on the available evidence. Mechanica) failure.--Flight control or trim malfunctions or flight insteumeut failures could account for the unwanted descent; however, the damage to and destruction of the flivht control components and flight instruments precluded a positive conelusicn that they were functioning properly. However, the Safety Board believes that such system malfunctions are improbable becuse the airplane was controllable until a few seconds before impact, the components examined revealed no preimpact failures, and there was no previous record of malfunctions. The evidence indicates that the pilot probably recognized the unwanted descent at the last moment and that he attempted to return the alrplane to climbing flight. The Initial ground contact Involved primarily the propellers cutting sagebrush until the airplane passed through the fence a few seconds later. The alrplane c auld not have had a high rate of descent at that time or there should have been extensive ground sears and gouges, and separation of airframe parts. The airplane remained in the air for about 450 feet after striking the fence, until the first principal ground contact was made. The only airplane parts in the initial contact area were the right-engine propeller dome and a piece of right-engine cowling. These pieces separated when the right engine hit the fenee post. Therefore, the Safety Board concludes that the airplana's descent was virtually nil, and the pilot had nearly recovered from the descent when the propellers touched the sagebrush and when the right engine struck the fence. The Safety Board believes that the reason for the unwanted descent was not @ problem that prevented ascent; rather, it apparently involved a mroblem that was not recognized in sufficient time to prevent the initial descent and to recever completely. Consequently, the Safety Board concludes that uncontrollable mechanical faflures probably were rot the reason for tha Joss of altitude. Runaway trim.--Alrplane design ecitecia specify that the pilot be able to he forces overcome ¢ Involved with runaway nosedown elevator trim. However, the possibility existe that runaway trim could have occurred concurrently with the eleplane configuration changes assoolated with the execution of the missed approach, and during a phase of flight In which the demands on a single pilot are very high.

[ANALYSIS] With poor visibility you may begin to go on instruments and then sensory illusions can make you delieve your instruments are wrong. There is just one way to beat false interpretation of motion. Put your faith in your instruments and not in your seuses. Know what kind of teicks your senses can play on you, keep caim, and have confidence in your instrument panel. 2. ANALYSIS 2.1 General The airplane was properly certificated and had been maintained In accordance with applicable regulations and procedures, The pilot was properly trained, certificated, and qualified to conduct the filght. He had the proper medical certification, adequate rest, and there was no evidence of preexisting or incapacitating disease or pathology which would have affected his ability to conduct the flight. Examination of the engines and propellers revealed that the engines were developing high power at impact. There was no evidence that power loss or malfunction were causal factors in this accident. There was no evidence found to indicate preimpact failure or malfunction of the airplane structure, syetems, or flight controls; however, severe Impact and postimpact fire damage destroyed many components which therefore could not be examined. There was no indication that the pilot encountered difficultles during the approach until the airplane passed the immediate vicinity of the VOR station (missed approach point). Therefore, there is no reason to suspect that airplane or ground navigational equipment malfunctions were causal factors, nor is there reason to suspect pitot-static malfunctions were factors. The fact that the eyewitnesses observed the airplane fly over the airport at or near the proper course and altitude further discounts such possibilitics. The witness observations and the fact that the landing geur and flaps were up suggest that the pilot was executing a missed approach procedure when the airplane crashed. Because the company procedure for a nonprecision approach specified gear extended, flaps at the approach setting (159, and airspeed 120 KEAS, the Safety Board concludes from the flap and gear positions that the pilot was unable to land from the approach and that he raised the gear and flaps at or near the VOR and Initiated the missed approach procedure. From the location of the ‘ed light deseribed by the surviving passenger, the Safety Board concludes that she saw the landing gear in-transit Nght which on when the gear doors sre open during landing gear retraction. The Safety Board concludes that the weather conditions at the time of the accident probably wero below the landing minimums specified for the approach being conducted. Byewitness accounts and official observations indicated deteriorating celling wh Oy a A : feo og 7 oe edge 4 el WT aig PERE PEE CME POR Tee ee a Sees ite dee ey = eee Re ee sb mean ola aceite -[4- and visibility about the time of the accident. The cloud ceiling decreased from 490 fect obscured at 2000 to 200 feet obscured at 2015, Also, the visibility decreased from 1 mile to 1/2 mile in moderate snow and fog during the same period.

[ANALYSIS] Runaway trim.--Alrplane design ecitecia specify that the pilot be able to he forces overcome ¢ Involved with runaway nosedown elevator trim. However, the possibility existe that runaway trim could have occurred concurrently with the eleplane configuration changes assoolated with the execution of the missed approach, and during a phase of flight In which the demands on a single pilot are very high. Consequently, a baat rey =e # de ae ee omen ee en en eee eee aE LOO SC nna SSE TTC ens SEO W On Unni . “ a ee i ia ys Na = . sae “a = on “ . > a aah = 5 eae ee ee ein re y Se . - ie : : ~ oo . es — a a - t cnn TIE ene ~ ame - nn , - t . runaway trim condition could have gore undetected for a sufficient length of time to place the airplane in a descending attitude that was not detected by the pilot In time to overpower the mistrim condition and to effect recovery from the descent. Impact and fire damage precluded positively eliminating the airplane trim system as a causal facter, and runaway trim could explain the reason for the accident although there {{s no history of this problem with PA-31 type airplanes. Autopilot use.--If the pilot was using the autopilot to fly the approach and to execute the missed approach, a malfanction or improper action by the pilot also could have initiated the descent and gone undetected for a short time. Although the compeny unwritten procedures and guidance specify that a pilot not use the autopilot for instrument approaches, the possibility exists that the pilot of N41070 did. The capab'' ; of the Century 41 autopilot to track the inbound course and to hold a desired alti during the approach would have reduced pilot workload. Similarly, the pilot could have readily executed the missed approach by means of inputs to the autopilot. Therefore, the Safety Board could not rule out the possibility that the pilot used the autopilot during the approach to reduce workload, or merely to see how well it performed. Moreover, the audio warning described ty the survivor could have been the autopilot disconnect alert. Although the survivor recalled hearing a "bell," there were no bell-type aural warnings installed in N41070. The sound of the autopilot disconnect aural warning is a bong-type sound which might have been the sound the survivor recalled hearing. If it was, it could mean that the pilot was using the autopilot during the approach and it was disconnected during the missed approach. The attopllot could have been disconnected when the pilot pushed the disconnect button, or when he overrode it by pulling back on the control yoke. {{ the autopilot malfunctioned and initlated the descent, or the pilot made an improper input once he recognized the deviation, he would have disconnected the autopilot eitiner by overciding it or pushing the disconnect.

[ANALYSIS] Such a maneuver could also explain the high angle of attack the aircraft assumed upon becoming airborne again. The comments of the crew indicate that a wind shear might have caused the short touchdown, or that the aircraft encountered a downdraft just after it crossed the chreshold. However, no evidence that the flight encountered either phenomenon was found. The surface winds reported by the tower never exceeded 10 knots. Although the firet officer referred to gusty winds during the approach segment. of the Flight, ‘the Board must conclude that these conditions did not exist at the runway threshold; indeed, he did amend his comment. “windy gusty" with the comment “Aw I mean windy out over the ocean there. . . .*# In addition to the possibility that it wa caused by a downdaft or gust, the short touchdown could be xplained by an faliure of the pilot to flare tha airccaft before touchdown. If the aircraft were flown ir such « manner that the pilot's eyes were held right on the ‘JASI _t1d@ slope all the way to touchdown, the main landing gear would touchdown quite near the point (300 to 365 feet beyond tie threshold) where Flight 505 did touchdowm. For example, calculations indicate that the main landing gear would contact the runway approximately 490 feet. short of the aiming point, or 400 fees. down this runway if the following representative conditions were assumed: distance from pilot aft to main danding gear of 70 feet; height of pilot's eyes above main landing gear tires of 14.8 feet; aircraft flown at 2.4° deck angle; anc at a deacent angle of 3.19. Another possible explanation for the short landing would be that the pilot performed a "duck-under" maneuver, This is a maneuver in which a pilot consciously positions his aircraft below the glide slope at a certain distance from: the runway threshold in order to permit an earlier flare to a landing, thereby giving himself more available runway on which to stop the aircraft. This maneuver, however, is inherently dangerous if not fully understood. The desceut below the original glide slope may require an appreciable increase in thrust to maintain the aircraft on a Nia OA Thoehge NAMIE RT pe alee ee et Reale a ec, oN AS, LN wt an ea po NE Gh tee meee Rm ase SOLS pa new and more shallow glide slopa to the desired touchdown point. If thrust is not increased, the aircraft will touchdown short of the desired touchdown point. Although the exact cause of the initial hard touchdown could not be determined, this landing did not cause catastrophic failure of the aixcraft, and it did not result in a subsequent uncontrollable maneuver. It is the opinion of the Board that, regardless of the physical or mental limitations imposed by the short runway and the surrounding hilly terrain, the pilot should have been able to recover from the bounce which followed the initial touchdown.

[ANALYSIS] In its analysis of the facts and circumstances of this accident, the Board assessed tie evidence bearing on the man-machine-environmental relationships. This approach led, in turn, to the formmlation of varicus hypotheses concerning the most probable causai areas of this accident. Tne first hypothesis considered ‘that @ destructively hard first touchdown occurred which was caused either by improper crew technigues or by external factors such as wind shear or turbulence. Another hypothesis considered that a mechanical failure which occurred sometime in the sequence of events was a direct cause of the accident. A final hypothesis is concerned with a breakdown in the interaction of the flightcrew with their aircraft subsequent to the initial touchdown. In the process of testing the3e hypotheses with observations made during the course of the investigation, the implication of the final hypothesis - the mnan/machine interface - became obvious. The factors which influenced the actions of the crew subsequent to the touchdown and the underlying factors prompted thes# events emerged as those of primary interest in determining the causal area of this accident. Before the third hypothesis is considered, the findings which disproved the first two hypotheses concerning possible cavaal areas will be discussed. Since the first causal axea presumes a hard initial touchdown as the direct cause of the accident, the nature of this landing must be reviewed. This first touchdown took place, according to witnesses, quite close to the end of the runway - approximately 300 feet beyond the threshold according to a controller in the tower, or 365 feet, if the tixe marks on the runway are accepted as those of this aircraft. This would place touchdown 435 to 500 feet prior te the VASI aiming point. The intensity of the touchdown was generally rated by surviving passengers as "hard", but not so hard as those following it. One witness described it as firm, ‘out not of an extreme nature. This witness was surprised at the height of the ascent that followed. In his statement, the captain described the landing as “very hard" and "very firm" - the flight engineer, as ". . . hard, definitely hard, but within safety bounds.% Tne accelevation trace of the fight data recorder confirms these Statements; the incremental accelerations recorded at the second and third touchdowns were beth approximately threa times that recorded at ths first touchiown. The physical evidence does not support the theory that the initial landing was catastrophic; the first evidence of scructural failure was located approximately 506 feet down the runway from the point of second touchdown. Finally, this theory is refuted by the lack of immediate concern shown by the crew. Only a few notuncommon remarks concerning the hard touchiown were made in the cockpit. It was not until slightly before the second touchdown when the voice record began to show a sense of {{mpending emergency in the tone cf voice and the comments made ky the crew.

[ANALYSIS] The evacuation, which wae accomplished within 1 minute, was well handled by the cabin attendants. The success of the evacuation is attributed in part to the fact that fuel spillage was minimal and that the aircraft used a fvel with a combustibility which retarded the immediate intensity of the rire. b. Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the captain's use of improper techniques in recovering from a high ro:nce generaced by a poorly executed approach and touchdown. Lack of <scochpit crew coordination during the approach and attempted recovery contributed to the accident. BY THE NATIONAL TRANSPORTATION SAFETY BOARD: Necember 29, 1971 /2/ /8/ /s/ /s/ /a/ JOHN H, REED Chairman OSCAR _M. LANREL Member LT SS LaLa emoer FRANCIS Il. MCADAMS r LOVIS M. THAYER Member ISABEL A, BURGESS Member All times are local, Atlantic standard time, based on the 24-hour clock unless otherwise specified. Very high frequency omnidirectional radio range. Visual approach slope indicator. A computed reference speed based upon 1.3 times the stall speed of the aircraft in the landing configuration with the engines at zero thrust. ‘This speed varies with the weight of the aircraft, and it may have speed increments added to allow for factors such as gusty winds. ag mee Kips (1,000 pounds) per square inch. D. Russell Davis, Department of Medicine, University of Cambridge -(1958) Ergonomics, 2.24. LG cl ' ois = 4 4 ry ae: Ee. * oa a . 5 *% me ac ‘ i A “¥ ry ¥ ' I fea “oe i: 1 y i i; t fy * ' : 1 a sy R. a, I _— « Uy I a go B= > d oni eK ¥ yl 5 i a ia > on ' 3 : illest esenatntenenntin tiene mnaninanener emmeenetiinemeatenieniaeanenemtaimetinmiiaiitiediddadn yt ttiilod aa APPENDIX A INVESTIGATION + , - I to : 4 . 3 owe t hed +. be 4 ; ' . ¢ F ‘ ’ 'y a ¥ 4 nc I if i I 7 ! t Lj 1. Juyvestigation The Board received notification of this accident at approximately 1400 e.8.t. on December 28, 1970. An investigating team departed from Washington, D. C., at 2107 that evening, and arrived at. the Hecry S Truman Airport, St.

[ANALYSIS] Although the exact cause of the initial hard touchdown could not be determined, this landing did not cause catastrophic failure of the aixcraft, and it did not result in a subsequent uncontrollable maneuver. It is the opinion of the Board that, regardless of the physical or mental limitations imposed by the short runway and the surrounding hilly terrain, the pilot should have been able to recover from the bounce which followed the initial touchdown. Based upon its examina:ion of all evidence, the Board concludes that the high bounce vas more the result of pilot input to the controls than of eiastic rebound of the aircraft as a result of a very hard landing. This conclusion stems from the following facts: neither crewmembers nor passengers felt that the initial landing was excessively hard; the aircraft attained 4 great height (50 feet) on the first bounce; and the flighc engineer stated that the aircraft assumed an excessive pitch attitude as it began the ascent. Once this bounce occurred, the captain had two choices of action according to the Boeing Training Manual: (1) he could have completed the landing by performing the high bounce recovery technique; or (2) he could have executed a goraround to make a gecond approach. He attempted to Salvage the landing. However, possibly because of limitations imposed by the shoxt runway and the surrounding hilis, he modified the bounce recovery technique in «the following manner: (1) additional thrust was not applied to lessen the rate of descent during bounce recovery; and (2) the wing fiight spoilers were deployed while the aircraft was airborne. The timing of the spoiler actuation appears at first to he debatable. The flight engineer stated that he observed tha captain actuate the spoilers shortly after the aircraft crested the first: bounce. However, the initiation of the intermittent horn sound on the cockpit voice record did not occur wntil after the second touchdown. This horn ie the warning signal of an unsafe flight condition which is believed to have occurred because the spoiler ileve. was moved from the 0° detent while the flaps were extended. Since the timing of the horn and the flight angineer's statement seemed contradictory regarding the timing of spoiler actuation, the Board investigated further to resolve thig macter. The cause of this discrepancy appears to be the result of a 2-1/2 second delay between actuation of the spoiler lever and the initiation of the warning systenr. This delay was observed by a Board investigator in several similar model 727 aircraft. A delay of this magnitude would then place the actuation of the spoiler lever ac a time 1 second betore the circraft touched down the second time. fhe physical data also agree with that interpretacion. The impact of the tail skid at, or immediately after the second touchdown, indicates that the aircraft was operating at a high angle of attack at that tine.

[ANALYSIS] Finally, this theory is refuted by the lack of immediate concern shown by the crew. Only a few notuncommon remarks concerning the hard touchiown were made in the cockpit. It was not until slightly before the second touchdown when the voice record began to show a sense of {{mpending emergency in the tone cf voice and the comments made ky the crew. Based upon the preceding evidence, the Board concludes that the initial touchdown was not of a destructively hard nature. The next possibility explored was that some malfunction of $$the aircraft caused this accident. Malfunctions which might possibly have been involved include loss of thrust, control system malfunction, landing gear malfunction, and pilot seat failure. The first two ct ee at GOREN SEE 23 te cine ora) malfunctions may be dismissed summarily since the crew did not report any problens in these areas and since no evidence of such malfunctions was observed in the Board's examination of the wreckage or of the flight recorder data. The possibility that a malfunction of the right main landing geax (RMLG) caused this accident precipitated an extensive study of that system. This study produced no evieence that any preimpact malfunction existed in the landing gear or its attach structures; rather, it demonstrated that the parts examined were sound, and that all fractures were caused by overloczrds applied to the RMLG. The pxobable failuice sequence advanced by the Boeing Company seems reasonable to the Board. The second touchdown, which was described as the hardest of the three, overstressed the year attach structure and the failure was jnitiated at that time. Passengers, it should be noted, described grinding sounds, and some thought that something on the RMLG broke at that time. The aircraft then apparently became airborne, after the FS 940 frame began to separate from the fuselage skin, but before the separation was complete. The pieces of fasteners reportedly found about 2,000 feet down the runway are consistent with this theory. The third touchdown then completed the failure of the gear attach structure. . This final disruption of its attach structure then allowed the landing gear to displaced upward until the right wing tip and the trailing edge of the outer wing flaps began to drag on the runway. Having thus determined that the environmental and machine elements were not likely causal factors of this accident, we now turn to the man and man/machine aspects of the operation, A study of the probable sequence of events which occurred during the approach and landing,and the factors which influenced those events, will show more clearly the involvement of the crew in the causal areas of the accident. In this respect, certain aspects of the approach of Flight 505 seem noteworthy. « 27 - One such aspect is the somewhat reversed student ~ instructor relationship which developed in the cockpit @uring the approach. ‘This relationsnip was evident in the decision of the captain to experience the rasponse of the aircraft at slower speeds while on final approach. The factors influencing this decision will be discussed later in this report.

[ANALYSIS] It was, perhaps, his preoccupation with these aspects of the approach maneuver which caused the captain to fiy the aircraft into a situation from which a short and hard landing was inevitable. Finding hingelf in a difficult situation immediately after the touchdow, seems to have confused the captain and affected hie further actions to salvage the landing. The Board feele that two factors may have combined to cause this .esponne s (1) the captain's lack of familiarity with the characteristics of the aircraft, in conjunction with the limitations imposed by the short runway at the Truman Airport, made him uncertain of the corrective action required; and (2) his power of reasoning was disrupted by natural behavioral changes which can occur in situations guch as that with which he was faced. The second factor concerns an individnal's natural respense to dangerous situations. Experiments conducted by Davis and reported in his study “Human Ferrers and Transport Accidents" 6/ explain the nature of this /cesponse. Davis noted that =man, like all animals, tndergoes certain behavioral changes) when danger appears imminent. These changes are intended to extract him rapidijy and impulsively from shat dangerous situation without having to go throug’ a slower reasoning process. In experiments in an artificial cockpit, Davis showed that this so-called emergency mechanism is detrimental in a situation which requires deliberate responses because it cancels the functioning of reasoning. These experiments showed that when a pecson reacts toward a situation in a way that experience (and training) have taught him to be effective, and that specific reaction deteriorates the situation instead, the emergency mechanism may set in within seconds. This creates confusion, which in turn, increases the sense of danger. A vicious circle is then formed which leads eitrer tc total inaction or to fruitless measures. In relating this theory to the circumstances at hand, it is intexesting to note that the captain's ~ttempts to salvage the landing by certain actions (such as an abrupt change in pitch attitude in the first place, and then by actuation of the spoilers during the bounce) only caused the ee ee OM Loe Ee ONT TE Tee ee Fe Te CE Eee er ee SE CE ee ere Oe A situation to deteriorate, contrary to his expectations. From that point on, che actions taken by the captain do not efem to be entirely rational. b. Post Crash Aspacts (1) Survivability Based upon the most common means of xneasurement, this was a survivable accident: The fuselage remained relatively intact; most of the occupants remaincd restiained; and the occupants had various means of immediate escape from the post-impact fire.

[ANALYSIS] In this respect, certain aspects of the approach of Flight 505 seem noteworthy. « 27 - One such aspect is the somewhat reversed student ~ instructor relationship which developed in the cockpit @uring the approach. ‘This relationsnip was evident in the decision of the captain to experience the rasponse of the aircraft at slower speeds while on final approach. The factors influencing this decision will be discussed later in this report. Another noteworthy aspect of the approach was the protile fiown. Although the heading was flown rather precisely, the indicated airspeed was never really stabilized. The airspeed underwent a short-cycle variation of 5 knots above and 3 below reference sped during the latter portion of the approach. Throughout the first half of cne 3-minute final approach, the sink rate? was approximately 600 fpm. This increased to a relatively steady rate of descent of 680 fpm during the final 50 seconds. That rate of descent, in conjunction with the average airspeed during that period, corresponded to an average descent angle of 3.1°. The first officer noted that the sink rate was 700 fpm just 4.4 seconds before touchdown, and the flight data recorder: readout shows a constant vate Gascent down to its “sero" altitude. The Board does not consider the error in the VASI landing aid a significant factor of this accident. It should not have caused the crew to fly an approach angle steeper than the misaligned VASI setting of 2.75°. However, as has been noted, the aireraft actually flew a descent angle during the last portion of the approach which was significantly greater than that projected by the VASI aystem. The initial teuchdown must certainly be considered noteworthy. The Board did not arrive at any positive determination of the cause of this. Among the various causes considered possible were: a late fiare, an encounter with wind shear or turbulence, failure of the pilot to fiare af all, and performance of a "duck-under" maneuver kv the pilot. The possibility that the “ircraft was flared too late to arrest «he rate of descé... before touchdown seems j i ‘ i {{ ee plausible, especially in view of the stuep approach angle flown in this instance. Such a maneuver could also explain the high angle of attack the aircraft assumed upon becoming airborne again. The comments of the crew indicate that a wind shear might have caused the short touchdown, or that the aircraft encountered a downdraft just after it crossed the chreshold. However, no evidence that the flight encountered either phenomenon was found. The surface winds reported by the tower never exceeded 10 knots.

[FINDINGS] See appendix B for these recommendations. 56 3. CONCLUSIONS Findings I. The landing gear, brakes, and tires were in good condition and were functional at touchdown. The airplane was maintained according to Federal Aviation Regulations, with the exception of anomalous seat belt attachment methodology and hardware, and there was no evidence of any systems or powerplant malfunction that might have contributed to the accident. The captain was properly certificated and operationally qualified for the flight in accordance with company procedures and Federal Aviation Regulations. Alr traffic control handling of the flight was appropriate. The captain had ILS glideslope data available during the approach but did not fly the ILS glideslope. If he had used the ILS, he would have been better able to assess the touchdown point. The tailwind during the descent, approach, and landing required a higher descent rate and resulted in a higher groundspeed at touchdown than that required if there had been a headwind or no wind. An altemate runway selection to provide a headwind component for landing would have been preferred. The partial obscuration of the airport environment, due to ground fog, contributed to the captain’s failure to recognize that the airplane was high on both his approach to the airport and subsequent landing attempt. The captain continued his attempt to land in the partial obscuration conditions, although reduced forward visibility 57 restricted his ability to determine the length of runway remaining. The PAPI for runway 6 would not operate ai its highest setting. After the tower closed at 2230, the pilot became responsible for determining the existing visibility and wind conditions at the airport. Both engines were operating at low or idle power at impact with the blast fence. The destruction of N990RA and the resulting occupant injuries were a direct result of the collision with the blast fence. Crash forces resulting from N990RA’s impact with the blast fence were survivable; however, the immediate postcrash fire created nonsurvivable conditions for occupants who remained in the front of the cabin. FAA interaction and comnuunication with local communities, althcugh persistent, were unsuccessful in gaining support for runway safeiy area improvements and for the installation of approach lighting for runway 6. Two of the four prior accidents at Sikorsky Memorial Airport in the past 10 years might have been avoided if approach lighting had been installed and had been operating on runway 6. The passengers and captain died as a result of smoke inhalation and/or thermal injuries. The passenger seats had been improperly assembled by Harrington Industries using unapproved parts, and seat belts had been installed incorrectly. 3.2 Probable Cause The National Transportation Safety Board determines that the probable causes of this accident were the failure of the captain to use the available ILS glideslope, his failure to execute a go-around when the conditions were not suitable for landing, and his failure to land the airplane on the runway at a point sufficient to allow for a safe stopping distance; the fatalities were caused by the presence of the nonfrangible blast fence and the absence of a safety area at the end of the runway. 60

[ANALYSIS] Further, the absence of any approach lights or other visual aids for the runway of intended landing could tend to keep a prudent pilot higher rather than on '® Class D is defined as that airspace from the surface to 2,500 feet above the airport elevation surrounding those airports that have an operational control tower. Two-way radio communication must be established with the ATC services prior to entry, and those communications must be maintained thereafter while in the class D airspace. There are no separatic” s. vices provided to VFR airplanes. Class E is controlled airspace that has no defined vertical urait but rather it extends upward from either the surface or a designated altitude to the overlying or adjacent controlled airspace. 49 glidepath during a visual descent to landing. The fog might have obscured the PAPI due to the reported failure of the high intensity light position. In that case, any visual flightpath information might not have been observed by the captain and might have contributed to a higher than normal approach path. The survivor stated that the captain pointed out the landing light switch on the overhead panel and told him, "when I tell you to, turn it on." At some point in the approach, the captain instructed the survivor to "tum it on." When the light was tumed on, according to the survivor, "it was like a strobe light." The captain then quickly said, "turn it off.” Estimates of the thickness of the fog layer varied. The survivor estimated the fog to be about 10 feet thick; the Learjet copilot estimated it to be between 15 and 20 feet; and airport personne! stated that it was 20 to 30 feet or more. The Safety Board believes that the landing light might have been tumed on during a critical phase of the landing as the airplane entered the reduced visibility area. The "whiteout effect” might have caused the captain to temporarily lose sight of the runway due to the light. The Safety Board also believes that the lack of visual cues for landing could have caused the captain to lower his rate of descent while trying to visually reacquire the runway, when he should have been initiating a go-around maneuver. The survivor stated that the captain was moving the contro] column as though he was "fishing" for the runway. After landing, the braking was hard before impact, according to the survivor. The survivor stated that ne could see the runway markings. The Safety Board believes that as a nonflying passenger in the copilot seat, he was most likely looking down and out the side window, and was therefore able to see markings on the runway. The captain, on the other hand, probably would be looking forward and out the front windscreen for landing, and therefore would have had a greater slant range viewing distance to the runway with increased obscuration from fog. If the captain was, in fact, “fishing” for the runway, this suggests that he was satisfied with the airplane’s alignment with the runway, and that he believed he was only slightly above the paved surface. Initial touchdown was not on runway centerline, but was on the paved surface with immediate correction toward centerline during rollout.

[ANALYSIS] Also, numerous seat belt installation errors were attributed to the company that refurbished the accident airplane and other airplanes examined during the investigation. Analysis of available weather data indicated that at the time of the accident, the surface conditions at the airport most likely consisted of a totally obscured sky, prevailing visibility of less than 1/4 mile in fog, and surface winds about 250 degrees at 4 knots. The tailwind component on final approach was about 21 knots at 3,000 feet, around 20 knots at 2,000 feet, and diminishing to around 4 knots at the runway. The emergency response was timely and efficient. 2.2 The Appropriateness of the Approach The Safety Board believes that the captain, from a decisionmaking standpoint, certainly wanted to land at Bridgeport and deplane the BDR-bound passengers who lived in that vicinity. A failure to land at Bridgeport would obviously inconvenience the passengers and Action Airlines because it would 45 require deplaning the passengers at an altemate airport late at night and then providing ground transportation for their retum to Bridgeport. Weather information for Bridgeport that the pilot had received before departure was favorable, and the weather along the route was clear. Therefore, there was no reason to delay takeoff from Atlantic City or to plan to fly directly to an alternate airport. The captain would have observed the presence of fog as he approached the airport, but, from altitude, he should have been able to see through it to the ground and to observe major features of the airport and runways. He was unaware of the actions of previous pilots who chose not to land. Because he tuned to the New York approach frequency responsible for Bridgeport at 2246, he would have heard no communications with the pilots of the Sabreliner or Turbo Commander airplanes that diverted earlier. However, he would have been able to hear communications with the Learjet. At 2248, the controller asked the Learjet whether he had the Bridgeport airport in sight, and the Learjet reported “we've got the runway in sigiit now." At 2251, Action Air 990 changed radio frequencies to the Bridgeport common frequency. This frequency was not recorded, and the Safety Board has no direct record of what the pilot said or heard. However, the Learjet pilots indicated that they received a radio transmission from the accident pilot inquiring “how's the weather down there?" When he received no answer, the pilot inquired "how was the ground fog?” and was advised by the Learjet crew, "Not bad until you get on the ground.” The surviving passenger corroborated this by stating that the Action Air captain received weather information from the Learjet crew indicating only a thin fog. The Safety Board believes that it was appropriate that the pilot radioed ahead for weather information from the Learjet. Although he had a weather forecast and could make his own observations, the most credible and timely observations of the severity of the fog would be those made by a flightcrew that had just landed.

[PROBABLE CAUSE] Safety issues in this report focused on the instrument landing system, runway safety areas, and runway lighting systems. Safety recommendations conceming these issues were made to the Federal Aviation Administration, the Connecticut Department of Transportation, the City of Bridgeport, and the Town of Straiford, Connecticut. Also, as a result of the investigation of this accident, on May 11, 1994, the Safety Board issued Urgent Action Safety Recommendations A94-111 and A-94-112 to the Federal Aviation Administration conceming aircraft maintenance performed by Harrington Industries, Inc. NATIONAL TRANSPORTATION SAFETY BOARD WASHINGTON, D.C. 20594 AIRCRAFT ACCIDENT REPORT IMPACT WITH BLAST FENCE UPON LANDING ROLLOUT ACTION AIR CHARTERS FLIGHT 990 PIPER PA-31-350, N990RA STRATFORD, CONNECTICUT APRIL 27, 1994 1. FACTUAL INFORMATION 1.1 History of the Flight On April 27, 1994, about 2256 eastem daylight time (EDT),’ .. on Air Charters flight 990, a Piper PA-31-350 Navajo Chieftain, N990RA, crashed into a blast fence at the departure end of runway 6 after landing at Sikorsky Memorial Airport (BDR), Stratford, Connecticut. The flight originated in Atlantic City, New Jersey, and operated on a visual flight rules (VFR) flight plan. The airplane was operating under Title 14 Code of Federal Regulations (CFR) Pan 135 as a single pilot, on-demand passenger air carner flight. Following landing touchdown and during rollout, the airplane collided with a blast deflector fence at the departure end of ninway 6. The airplane was destroyed by impact forces and a postcrash fire. Eight of the nine occupants sustained fatal injuries. One passenger was seriously injured. The chartered series of flights began on the morning of Apni 27, 1994. The flights were contracted by an independent representative of Resorts Intemational, Inc., a gambling casino operator located in Atlantic City, New Jersey. The flights operated as Action Air flight 990 from Hartford/Brainard Airport (HFD), Hartford, Connecticut, to Pomona Airport (ACY), Atlantic City, New Jersey, with an intermediate stop at Sikorsky Memorial Airport. A return flight was scheduled in the evening to return the passengers to their potuts of origin. ‘All times in this report are in castem daylight time, using the 24-hour clock, unless othemvise The captain reported for duty at 0800. The airplane was fucled with 116 gallons of fuel, as recorded on the airplane weight, balance, and load manifest. The flight was scheduled to depart HIF& at 0830 with two passengers and one company employee; however, the flight was delayed because the BDR weather was below landing minimums.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] Thus, as the airspeed was reduced during the initiation of the instrument approach. the “stick shaker" would have been activated when the indicated airspeed reached 145 m.p.h.. and would have alerted the crew to an unusual situation, Further, as the aircraft descended, there would have been an increase in indicated airspeed wichout any change in power settings, or without a change in the pitch attitude of the aircraft. The Board believes that all of these indicators would not have lwen missed or ignored by both the pilot and copilot, but particularly by the pilot, who had an unusually high degree of skill according to pilots who had flown with him. Accordingly. the possibility of an altimeter error as the result of static system restriction is rejected. ad. The remaining possibility is that the wescent was intentional, that the pilot was preceeding by means of visual reference to ground objects, bet, because of the restricted visibility wid rain droplets on the windsereen, he did not see the powerlines in time to avoid collision. Because of many variables such as actual descent rates, airspeeds, variance in ground speed duc to wind, actual altitude and time at departure from the VORTAC, and possible mancuvers by the pilot in navigating the aircraft, the exact time and geographic point at which the aircraft arrived at the clevation of the powerlines could not be determined. Nes, because of the extensive damage to the aircraft during wire and ground contact, and the subsequent fire, was it possible to deterraine the exact attitude or heading of the aircratt immediately before collision. However, the witness who heard the aircraft over her house was locared only 400 feet south of the centerline of tie -Peoria VORTAC 095° radial. Accordingly, it is believed that CSO 804 proceeded inbound along the VOR Runway 12 final approach course, with the possible exception of a last minute maneuver in an attempt to avoid the powerlines. fz is believed that the aircraft was operated beiow the MDA mtentionally, i order to make an approach to Runway 4 by neans of ground visual reference, for the following reasons: (1) The pilot had been informed that it was possible to “fy right around the airport and keep it in sight’ at 2,000 feet.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] That the powerplants were producing power at impact is evident from metal spatter on the turbine blades, the 45° torsional twisting of the exhaust casings, and propeller blade deformeation. The midrange power setting of the engines, Powe disclosed by the propelier blade angles at impact, was sufficient to maintain the aircraft in level flight at speeds in excess of 140 m.p.h.,'° or 50 m.up.h. above stall speed. Substantial forward velocity at the tine of initial contace with the powerlines is indicated by the following: (a) two stecl reinforced cables 0.883-inch in diameter, cach with a tensile strength of 22,500 pounds per square inch, were severed successively, and a third cable was damaged, (b) the cables destroyed the aircraft’s windscreen and severed +’ the right wing, (c}} the left wingtip struck wbground 75 feet beyond the point of imeros contact with the powerlines, which caused the wing to bend at midspan and separate from the fuselage at impact, vet the forward momentum was great enough to cause the airframe to strike the tree 152 feet beyond the point of initial contact with the powerlines, and (d) the velocity at impact with the tree was still great enough to cause the fuselage to split apart, and light debris aad two bodies to be thrown approximately 735 feet beyond the tree along the wreckage path. Accordingly, the Safety Board concludes that the aircraft was fully controllable prior to impact with the powerlines, and that descent below the prescribed minimuin safe altitude was not duc to mechanical problems with the aircraft or incapacitation of the ccewmembers, Reasons other than mechanical failures or operational emergencies which could result in flight below the MDA include: missetting or misreading of the altimeters; misreading the instrument approach chart; malfunction of the altimeters, or restriction in the Pitot static system; failure of the crew to monitor altitude during the descent; and, an intentional descent below the MDA in an attempt to establish and naintain visual reference to the ground, Each of these possibilities was considered in light of the information developed during the investigation CO ee ed np - miles per hour, The airspeed indicators ta NSICS were calibrated in both knots and miles per hour. M.p.h. was used operationally. ve at oe tw dk wi ee Swe * ov Su a eR” eter PS bh s Saent ean i and subsequent public hearing. All but the possibility of an intentional descent were rejzcted for the follesving reasons: a, Missetting of the altimeters.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] Norwithstanding che first Ozark Air Lines flight had been unsuccessful in its attempt to fand because of the inability to keep the airport in’ sight while it atcempted to line up with Runway 4. (2) The crew of the second Ozark fight testified that occasional visual reference to the ground was established as the aircraft proceeded inbound from the VORTAC to the airport. (3) The pilot of CSO 804 was exceptionally skillful and intimately familiar with che airport environment. (4) The reported ceiling at the airport was 80 feet below the MDA, A “duck under” maneuver to position the aircraft below the MDA would be necessary if ground reference was to be established at a distance chat would permit aligning the aircraft with Runway 4° for landing. Conversely, if the approach was made in accordance with prescribed procccures, and the aircraft was kept at the MDA until the environment associated with the approach end of Runway 4 was sighted, the pilot could anticipate a missed approach for the same reasons the two Ozark Air Lines flights ahead of hit had missed their a yproaches. Under these circumstances, it is likely chat the pilot would conclude that the essential prob lem of establishing and maintaining alignmeni with Runway 4 during the final approach to landing could be solved by an carly descent below the cloud cover, and the attainment of visual ground reference, It is apparent chat a deviation from the scandard approach procedure must have be. initiated before the VOR was reached, or shortly thereafter, in order for the aircraft to have descended to the 746-foot altitude at which it struck the powerlines. Either the aircraft was not at the specifizd 1,800 fect must. over the VORTAC, or an average rate of descent in excess of 1,000 fe pre minute was established after it passed «s VORTAC and continued through the MDA until shortly before impact. (See Appendix F.) The Safety Board believes that the greater probability is thar the crew of CSO 804 estab. lished momentary ground contact, shortly after passing the VORTAC, in the same manner as did the crow of the Ozark Air Lines flight preceding them.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] A descent was initiated during encounter with one of these breaks in the cloud layer, and continued datil the aircraft was below the lowest clouds, Beeause some of the scattered clouds in the lowest layer had bases only 100 feet above che yround in the vicinity of the accident site, it would have been necessary for the pilot to descend to an altitude slightly lowe: in order to maintain visual reference to the ground. Visibility ranged only from 1/4 to 1/2 mile in the area. From the aircraft cockpit it was probably somewhat less because of moisture on the windscreen. Accordingly, the absence of a well-defined horiton and the scarcity of geometric shapes on the ground below the aircraft's flightpath would have caused the pilot co direct his attention downward, as well as forward, and would have made the detection of wires against alow contrast background extremely difficult, if not impossible. The factors which could explain why the piloc attempted this course of action were: a. If the landing at Peoria were abandoned, the company would have had to pay the transportation costs of three passengers from an alternate airport back to Peoria. b. A crew chinge was to take place at Peoria, and if the landing was not effected, the pilot, the president of the airline, would have had to continue to fly the schedule, and would have been unable to attend to his duties as president of the airline. Since he was known to run every facet of the company business himself, without delegating authority to others, it is likely that he considered his daily presence at the company headquarters vital to the success of the airline. During testimony at the Safety Board's public hearing, a question arose as to the adequacy of FAA's surveillance and enforcement regarding compliance with Fedcral Aviation Regulations by Chicago and Southern Airlines, Hic. Three violations of these regulations occurred 16 montis prior to the accident, for which civil penalties had been assessed. The company continued violations of the crew flight time limitations and evidence of this was hidden deliberately from the FAA Gperational Inspector, The Safety Board believes that more aguressive followup should have been taken by the FAA General Aviation District Office having jurisdiction over Chicago and Southern Airlines, to insure that company’s continued compliance with all Federal Aviation Regulations. Company adherence to all Federal Aviation Regulations and the FAA surveillance and enforcement thereaf will be 2 special subject of the Safety Board’s forthconiing in-depth study of the overall air taxi operations. 2.2 Conclusions a.

[FINDINGS] Pieces of the wheel rim from either the No. 1 or the No. 2 wheel hit the tire and caused it to blow cut. This blow out affected further the aircraft's braking capability. Also, the left main landing gear might not have collapsed if No. 5 tire had been available to distribute load on the overrun area. The tires on the aircrvaft may have been operaied in the overdeflected condition, since the average inflation pressure was less than the optimum pressure for maximum gross weight. The aircraft left the departure end of runway 6R at a speed of about 68 kns. The aircraft slid to a stop about 83 sec after the start of the takeoff. It came to rest about 664 ft beyond the departure end of runway 6R on a heading of 008°. The airceaft could not be stopped on the available runway because of the partial loss of breking effectiveness attributed to failed tires and a wet runway surface. Dynamic hydiuplaning conditions were not present. Runway 6R had acceptable friction characteristics according to current FAA suggested criteria for the Mu meter; however, the Mu meter data could not be used to estimate aircraft stopping performance. During the 4-year period between the grooving of runway 6R/24L and the day of the accident, the airport operator did not make “he friction surveys suggested by the FAA. The FAA and the airport operators did not have ready access to equipment or trained personnel required to conduct periodic friction surveys. No FAA procedures or data are available to aircraft operators or flightcrew to relate degraded runway friction conditions to changes in allowable aircraft takeoff weights, decision speeds, and stopping distance. -~4l - The current FAA rejected takeoff requirements for aircraft certification, aircraft operations, and pilot training do not address wet ruuway, slippery runway, or tire failure conditions. It was not possible to determine accurately from performance analyses if the full braking capability of the sircraft was achieved during the initial phase of the rejected takeoff. In its 1977 report on rejected takeoffs, the FAA concluded that aircraft safety could be improved by accounting for wet/ slippery runway conditions and tire improvements. Flightcrew simulator training for rejected takeoffs ie inadequate because of the lack of FAA requirements for wet runway considerations in those simulators and for rejected takeoff training at the maximum takeoff gross weights and decision speeds encountered in normal operations. The landing gear attachment structure failed and caused the left wing fuel tank to rupture. Fire may have started before the aircraft left the runway surface. The evacuation was started promptly and elmost simultaneousiy throughout the cabin. The 1L exit was opened with the slide/raft handle in the disarm position. Slide/rafts at exits 2L, 3L, and 4L burned inomediateiy after they were deployed. All slide/rafts on the right side were deployed and used. The overwing ramp for the 3R slide/raft malfunctioned.

[FINDINGS] Their imacdiate response and their initiative in seeking al.ernate escape routes when the normal routes were rendered useless, undoubtedly saved lives and decreased the number of injuries. 3. CONCLUSIONS Findings The crewmembers were certificated and qualified for the flight. The aircraft was certificated, equipped, and maintained in accordance with FAA requirements, except for the inoperative CVR. The runway was vet, but there was no standing water. Runway 6R was the only runway available for takeoff. Two 12,000-ft runways, the use of which could have made a successful rejected takeoff possible, were not available to wide body aircraft. Lineup for takeoff began about 166 ft, from the approach end of runway 6R. The flightcrew used the minimim lineup distance and established takeoff thrust as required by company procedures. The captain promptly rejecced the takeoff at or below 152 kns (V1 speed was 156 kns) after hearing a loud "metallic bang" and feeling a “quivering" of the aircraft. The captain responded to the emergency by first applying brakes and then applying maximum reverse thrust on all engines. Ground spoilers actuated whei thrust levers were moved to the reverse thrust positions. Reverse thrust began about 5.8 sec after V1 was reached and peaked 3 to 8 sec after the engines began to spool up for reverse thrust. Reverse thrust was maintained above 100 percent Ny on ali three engines during the reversal sequence. Reverse thrust was maintained on the center and the right engine until just before the aircraft stopped beyond the end of the runway. Reverse thrust on she left engine ceased when that engine was torn from the aircraft, 100 ft beyond the end of the runway. The first tire feiled at the No. 2 tire position about 6,300 Ft from the departure end of runway 6R. The tire failed because of a thrown tread. The carcass blew about 4,520 ft from the departure end of the runway. The second tire failed at the No. 1 tire position about 4,480 ft from the departure end of rvnway 6R. Fatigue in the ply structure may have been caused ty long-term overload since the tire wis mounted on an axle with a tire of a different brand whic) had less sfdewall stiffness. The tire blew out because of an overload. The third tire failed at No. 5 tire position about 3,400 ft from the departur2 end of runway 62. Pieces of the wheel rim from either the No. 1 or the No. 2 wheel hit the tire and caused it to blow cut. This blow out affected further the aircraft's braking capability. Also, the left main landing gear might not have collapsed if No. 5 tire had been available to distribute load on the overrun area.

[ANALYSIS] However, they are presently studying the matter. Certification and operations regulations do not take into account the longer stopping distances required by rejected takeoffs on wet or slippery runway surfaces or the reasons for rejected takeoffs other than an engine failure. The FAA reached this same conclusion in its 1977 Jet Transport Rejected Takeoff Study. However, the FAA still he3 not developed ‘rocedures which would allow aircraft manufacturers, airline operators, or flight crewmembers to determine changes in decision speeds or aircraft gross weights, or both, so that successful rejected takeoffs could be accomplished from near V1 peed on a wet runway following engine or tire failures. The Safety Board c ...curs with the recommendations in the FAA's study and believes that, unless FAA takes wet or slippery runways and the other reasons for rejected takeoffs into account, accidents and incidents will continue, especially when aircraft are required to operate on dry, wet, or slippery runways of critical length. Flightcrews, for the most part, are trained for rejected takeoffs ta flight simulators, and vherefore, training is limited by the capability of the simulator and by the training requirements of 14 CFR 121. Since the simnulator's accelerate-stop performance is based on the aircraft / / cA manufacturer's dry runway, engine-out fertification data, it is impossible to simulate realistic wet runway conditions or malfunctions other than an engine failure. FAA simulator rgquirements contain performance specif{{icaticns for the acceleratios{{ portion of the rejected takeoff maneuver, but nct for the stopping portion of the waneuver. Further, the FAA does not requfre that eye instructor, evaluation pilot, or trainee determine pilct reactich time and the amount of braking effort ‘applied by pilots in the simusator. The training reqy rements of 14 CFR 121 do not require rejected takeoffs at the maximum gross weights and decision speeds encountered in normal operations. As revealed in testimony at the public hearing, crewmembers do not typigally receive this more demanding training. Although a captain is expected to know that a maximum braking effort would be required when a rejected takeoff is initiated at the higher speeds and gross wejghts, he cannot be expected to judge whether the aircraft is decelerating at its maximum capability if he has never been trained for that eventuality. The captain of Flight 603 reacted promptly to the tire failures, and he acted iff accordance with Continental procedures. During depositions, he stated that he applied full brake pressure immediately. However, because there was no requirement for the DFDR to record brake pressure at the brake or flight test data available to validate the Douglas/NASA predicted deceleration rates, the Safety Board was not able to verify if maximun brake pressure was achieved during the early porcion of the rejected takeoff. Further, the Safety Board could not determine if the antiekid system performed to its maximum capability during the sane time period.

[ANALYSIS] Furthermore, since the aircraft's performance data were obtained through testi ig on dry runways, there is no assurance that the current concept is adequate when the braking coefficient of friction is reduced on a wet surface. Even when full braking capability js available and the runway surface is dry, a rejected takeoff initiated at or just before the aircraft reaches Vj speed is risky on a minimum length runway. Using maximur braking and optimum procedures, the aircraft is going to use all of the remaining runway length to stop. (Actually, a smail margin is provided since the braking effect of thrust reveraal is not considered in acceleratestop performance data.) In this accident, the aircratt had about 800 feet more than the minimum runway provided by the balanced field concept; even so. ‘wo significant factcrs combined to invalidate the use of V1. speed €3 .. , »-no-go decision point: (1) the loss of effective braking on whee]: 7:th blown tires, and (2) the reduction in brake friction coefficient on wet surface. The Safety Borrd, therefore, views the captain's no-go decision as a key element in the accident sequence. This is in no way intended to imply that the Board faults his decision, but rather that the limited validity of the decision-making process and of the V}} concept in its entirety justifies further analysis. As the aircraft approaches the decision apered (V1) the decisionmaking time available to the pilot decreases. At Vj speed he has no time for a decision and must respond immediately to reject the takeoff {{f he is to be able to stop the aircraft on the runway even under ideal conditions. A failure to act promptly to any problem encountered as the aircraft approaches the Vj speed can have catastrophic results if the problem is an engine failure, is structural in nature, or is associated with loss of critical systems or fiight controls. Certainly, these possibilities are ever present in a pilot's mind when something unusual occurs at a critical time. Therefore, the dominant tendency is most likely to reject the takeoff when any unevaluated anomaly occurs before Vi speed, particularly if the problem is accompanied by noise and vibration. In this accident, the captain heard a loud metallic bang and the flight data recorder indicated that this occurred 1.2 seconds before the aircraft reached Vj speed. The captain was therefore faced with the need for immediate action. He had no time in which to evaluate the significance of the loud bang and vibration if he was to successfull: reject the takeoff. However, it became evident during the Board's investigation that the noise and vibration were associated with a tire failure and that the aircraft could undoubtedly have been flown cff the runway successfully.

[ANALYSIS] The captain was therefore faced with the need for immediate action. He had no time in which to evaluate the significance of the loud bang and vibration if he was to successfull: reject the takeoff. However, it became evident during the Board's investigation that the noise and vibration were associated with a tire failure and that the aircraft could undoubtedly have been flown cff the runway successfully. The Soeing Aircraft Company, in its flight manual, and the McDonnell-Dougilas Corporation, in a recent flightcrew newsletter, have erphasized the potentially dangerous; nature of the rejected takeoff maneuver at or just below Vj at a critical field length, and have izdvocated a better pilot understanding and appreciation of the rejected takeoff decision and the abnormal conditions leading to that decision. Based upon its analysis of this accident and others, the Safety Board agrees that pilot preparedness is essential, and concludes that the problem is sufficiently important and complex to warrant a thorough review and revision of the Vj concept. Ideally, with a more comprehensive V}} concept the pilot would be provided a decision speed from which he could be assured of the capability to stop the aircraft on the remaining runway regardless of the reason and circumstances for rejecting the takeoff. This would apply to wet runways as well as dry runways and would account for degraded braking capability due to common failures such as blown tires. A .mittedl;, such comprehensive criteria may aot be practically achievable. Nonetheless, the Board believes that some improvements to the present criteriu are needed if accidents such as this are to be prevented. Fuel Tank Rupture and Fire Shortly after the aircraft departed the right corner of the departure end of the runway, the left main landing gear broke through the macadam surface of the overrun area and failed. The increased footprint pressure exerted by the No. 6 tire and the load from the Nos. 1 ard 2 wheels was beyond the macadanm's support capability. When the DC-10 was certificated, the FAA required that the DC-10 landing gear attachment be designed so that, if it failed because of up and aft overload, no part of the landing gear structure could puncture any part of the fuselage fuel system. The manufacturer satisfied this requirement. In four other accidents, DC-10 landine gear have failed with no wing fuel tank rupture; however, in this case the loads imposed upon the left main landing gear exceeded design loads. As a result, -~ 37 ~- some of the landing gear attachment structure failed in an unusual mode and a large hole was torn in the aft web of the left wing rear spar at the juncture of the two left wing main fuel tanks. The fuel that was released through this rupture vas the major contributor to the extensive postcrash fire. The Safety Board was not able to determine conclusively where or when the fire started. Statements of some passengers and flight attendants indicated that fire may have been present in the area of the left main landing gear whcels before the aircraft left the runway surface.

[ANALYSIS] However, because there was no requirement for the DFDR to record brake pressure at the brake or flight test data available to validate the Douglas/NASA predicted deceleration rates, the Safety Board was not able to verify if maximun brake pressure was achieved during the early porcion of the rejected takeoff. Further, the Safety Board could not determine if the antiekid system performed to its maximum capability during the sane time period. After the captain and the other flight crewmembers became aware that they would not te able to stop the aircraft on the runway, the captain steered the aircraft to the right to avoid colliding with the approach light stanchions fcr runway 24L. The Safety Board believes that this action reduced the severity of this accident. Impact with the nonfrangible stanchions could have caused additional major structural © damage. The Rejected Takeoff Decision The determination of the minimum length of runway required for takeoff in air carrier operations, or conversely the determination of the maximum weight for the airplane to take off on any given runway, is based upon a balanced field concept. This concepe is predicated upon the calculated ability of the aircraft to either stop within the length of the runway or to successfully continue the takeoff after an engine failure during the takeoff roll. Before euch takcoff, the flightcrew will use accelerate-stop performance data obtained from the aircraft's certification tests to calculate the maximum allowable takeoff weight and the critical engine fullure speed (V)). The flightcrew has been trained to use the Vy speed as a decision point during the takeoff roll. If au engine failure ie recognized before the V) speed is reached, the pilot is trained to reject the takeoff and ne, in fact, must reject tne takeoff since he cannot be assured of successfully continuing. On the other hand, $f the aircraft is beyon?’ the V1 speed before ar engine failure is recognized, the takeoff must be centinued since the pilot cannot be assured of stopping the aircraft on the remaining runway. Although V) speed is designed to be che go-no-go decision speed in event of an engine failure, the Safety Beard believes that pilots have come to regard Vj}} as the go-no-go decision speed for any anomaly during the takeoff roll. However, the calculated Vj speed, by current definition and certification standards, is valid only fer circumstances in which the aircraft has its full braking capa>ility. Furthermore, since the aircraft's performance data were obtained through testi ig on dry runways, there is no assurance that the current concept is adequate when the braking coefficient of friction is reduced on a wet surface. Even when full braking capability js available and the runway surface is dry, a rejected takeoff initiated at or just before the aircraft reaches Vj speed is risky on a minimum length runway.

[ANALYSIS] Although other runways at the airport, which arg 2,000 ft longer than runway 6R, probably could have contained the rejécted takeoff, they were not available to aircraft with gross weight’ of more than 325,000 lbs because uf runway overpass strength limipations. A project to eliminate this limitation is in the planning stapes. The Safety Board urges the responsiole authcrities to expeaite this project and make longer, safer runways availabl. to heavier aircraft gt Los Angeles International Airport. . i ~ 7 - jf Even though te measured wet friction characteristics of , wunway 6R exceeded minimum ctandards suggested by the FAA, the Safety eens, ; TTL, » -~ 33- Board believes these characteristics contributed to the partial loss of the aircraft's brakirg capabilities and, therefore, contributed to the inabi’ity to stop the aircraft on the runway. This loss of runway friction was particuiar!y evident in the rubber coated areas on the departure end of the runway, the touchdown area for landings on runway 24L. The FAA developed the minimum runway friction standards and methods for the measurement of these ¢tandards. FAA Advisory Circular AC 150/5320-12, Methods for the Design, Construction and Maintenance o° Skid Resistant Airport Pavement Surface:, made this information available to airport operators. Because the inforration in this Advisory Circular is not mandatory, airport operators Jo not routinely use it. At the time of this accident, neither the Los Angeles International Airport cperat>r nor the FAA authorities in the Los Angeles area had the FAArecommended equipment to make these measurements, Furthermore, no record could be found to show that friction surveys had ever been conducted or that rubber deposits and other contaminents had ever been removed from the surface of runway 68/24L since the runway was grooved tn 1974. When the Safety Board made the results of its runway friction tests available, the affected areas were cleaned. For some time the Safety Board has maintained chat the provistons of AC 150/5320-12 should be made mandatory. As a result, the Safety ‘Board issued safety reconmendations A-76-136 and 137 on November 18, 1976. The FAA disagrees with the Safety Board, in that it believes the economic burden placed on individual airport operators would be prohibitive and the precision techniques for friction testing are not presently available. However, they are presently studying the matter. Certification and operations regulations do not take into account the longer stopping distances required by rejected takeoffs on wet or slippery runway surfaces or the reasons for rejected takeoffs other than an engine failure. The FAA reached this same conclusion in its 1977 Jet Transport Rejected Takeoff Study.

[FINDINGS] The evacuation was started promptly and elmost simultaneousiy throughout the cabin. The 1L exit was opened with the slide/raft handle in the disarm position. Slide/rafts at exits 2L, 3L, and 4L burned inomediateiy after they were deployed. All slide/rafts on the right side were deployed and used. The overwing ramp for the 3R slide/raft malfunctioned. The slide/raft at 1R failed from radiant heat danage; the girt bar supporting fabric failed at 4R because of overload or uneven load; all other slide/rafts burned. The evacuation was completed using the emergency rope which hung frem the first officer's side window. The first crash-fire-rescue unit was on the scene fighting the fire in about 90 sec from the initiation of the rejected takeoff. Two passengers died of burns and smoke inhaltion after exiting through the 3R exit. ~42- 36. Evacuation time was approximately 5 minutes. 3.2 Probable Cause The National Transportation Safety Board determined that the probable cause of the accident was the seqvential failere of two tires on the left main landing gear and the resultant failure of another tire on the same landing gear at a critical time during the takeoff roll. These failures resulted in the captain's decision to reject the takeoff. Contributing to the accident was the cumulative effect of the prctial loss of aircraft braking because of the failed tires and the recuced braking friction achievable on the wet runway surface which increased the accelerate-stop distance to a value greater than the available runway length. These factors prevented the captein from stopping the aircraft within the runway confines. The failure of the left main landing gear and the consequent rupture of the teft wing fuel tanks resulted in an intense fire which added to the severity of the accident. 4. SAFETY R:COMMENDATIONS As a result of this accident, the Safety Board, on September 6, 1978, recommended that the Federal Aviaiton Administration: “Assess current tire rating criteria, as used by the Tire & Rim Association and as interpreted by airframe designers and Federal Standards, in terms of compatibility of tire, airframe, and intended operation to assure that adequate margins are provided for all normal conditions. (Class II, Priority Action) (A-78-§7) “Upgrade Technical Standard Order C-62b to reflect current engineering practices and operational conditions in both the specifications for performance standards and certification test requirements. (Class I1, Priority Action) (A-78-68) "Insure that the tire is compatible with the airframe by considering this compatibility during the airplane certi‘acation.

[ANALYSIS] The onset of this wave depends on the ground-speed of the aircraft. -~32- In this instance, tire No. 2 weuld have then been overloadec. and overdefiected which could result in tread loss followed by sarcass blowout. Failure of the No. ? tire alone probably would not have affected the aircraft's accelerate-st » performance to the extent that an accident was inevitable. The Safety Board believes that had the No. 1 tire not euffered previous degradation, it would have been capable of operating for a longer period than evident in this accident. The examination of the tire's carcass disclosed advanced fatigue in the ply structure. In addition, there was evidence of severe cord overheating near the sidewall vead area, ari several other areas of the sidewall showed evidence of very nigh temperatures. Such corditions are typical of those produced by overload or overdeflected operation for a prolonged period of time. Although the Safety Board cannot determine when such damage was inflicted, ft is concerned that airframe and tire design, aad operational and maintenance procedures can combine to cause prolonged operation of tires In an overdefiected or overloaded condition. Normal differences between two tires on the same axle, particularly if they are of different designs, could preclude them from carrying equal loads. The Safety Board believes that the preexisting damage in the No. 1 tire was a facthr in causing it to ultimately fail almost immediately after the No. 2 tire failed, and thus, the preexisting damage may have been a causal taftor. Abbut 3,400 ft from the departure end of runway 6R, the No. 5 tire fail:43. "This failure was caused by foreign object damage when pieces of either the No. 1 wheei or No. 2 wheel broke off after the wheels contacted the runway surface and hit the No. 5 tire. This failure further reduced the braking capability of the aircraft. wv Rejected Takeoffs == “=> a Because of its gross weight of about 430,900 lbs, the only runway available to the aircraft at Los Angeles International Airport wae 4R, which was 10,285 ft long. Pased on current FAA dry runway certivication data, an 850-ft stopping margin would be expected if a reject.:1 ta off was initiated at Vy because of engine failure. However, wher et rutiway surface conditions and tire failures are considered, the stopping margin is eliminated. Although other runways at the airport, which arg 2,000 ft longer than runway 6R, probably could have contained the rejécted takeoff, they were not available to aircraft with gross weight’ of more than 325,000 lbs because uf runway overpass strength limipations. A project to eliminate this limitation is in the planning stapes.

[ANALYSIS] The Bresg weight and c.g. were within prescribed limits. The aircraft's airframe systems, and powerplants were not causal to this accident. Tne evidence showed that the accident was initiated by the nearly simultaneous carcass failures of the two tires mouated in the No. 1 and No. 2 positions. Since these tires were mounted on the same axle, che 97,920-ib load on the axle was distributed between the two tires. The analysis of tre and wheel marks on the runway indicated that the failure sequence began when the tread from the No. 2 tire separated from its carcass about 6,300 ft from the departure end of the runway. The tire carcass remained intact until the aircraft was about 4,520 ft from the runway departure end where squiggle marks indicated blowout. The squiggle marks on the runway at that point and portaccident examination of the tire remains indicated that extreme heat had built up in the carcass sidewall and that the carcass had blown out at its upper sidewall. After the treau had separated, the rubber was ~braded by direct contact with the runway surface and eventually blew out. After the No. 2 tire carcass blew out, the entire load on the axle was imposed upon the No. 1 tire. The markings made by the Nv. 1 rim, which showed contact with the runway surface 4,480 ft from the runway deperture end, indicated that the No. 1 tire failed aimost immediateiy (within two wheel revolutions) after the No. 2 tire carcass failed. The No. 1 tire virtually disintegrated while the whole No. 2 tire carcass, except the beads, came off the wheel. Examination of the remains of the No. 1 tire indicated that the tire ultimately blew out in the lower sidewall. The DFDR showed that the tires failed just before the aircraft accelerated through 152 kns--about 4 kns below the calculated V1 speed. The DFDR further showed that the captain reacted promptly to the tire failures and began rejected takeoff procedures. However, he was not able to stop the aircraft within the remaining runway. Thus, to understand this accident sequence, two distinct, but related, issues must be analyzed. First, since the tire failures triggered the sequence of events, tire failuree and tire reliability in general must be anaylzed. Second, reasons must be determined for the captain's inability to stop the aircraft on the runway even though the rejected takeoff was -nitiated before the aircraft reached V, speed. ‘Tire VaiJures and Tire Reliability Both the No. 1 and No. 2 tires were on their third retread cycle, a limit which was wet hv the airline based upon prior experience of unscheduled remova}} of DC-i0-lv tices. The two tires had been manufactured by differen: companies and had different design characteristics. Both, however, met all specifications sec forth in FAA regulation for certification.

[ANALYSIS] The fuel that was released through this rupture vas the major contributor to the extensive postcrash fire. The Safety Board was not able to determine conclusively where or when the fire started. Statements of some passengers and flight attendants indicated that fire may have been present in the area of the left main landing gear whcels before the aircraft left the runway surface. The escape of hydraulic fluid under pressure from ruptured brake and antiskid hydraulic lines 10/, and the friction heat developed from rubber and setal contact with tr~ runway surface could have ignited a fire. Fire engulfed the left side of the aircraft immediately after the left main landing gear failed. This fire continued until extinguished by the Los Angeles Airport Fire Department. The Safety Board believes that the quick response of the Los Angeles Fire Department, particularly fire Station 80N, prevented greater loss of life and lessened injuries to evacuees. This quick response was possible because the authorities at Los Angeles International Airport, unabJe to meet the required emergency response times, constructed auxiliary fire stations at the midpoint of the airport's 2 major runway complexes. The decision of the firefighters on CB-1 to position themselves so that firefighting agent could be used to keep escape lanes open for the evacuation wat exemplary and reduced the number of deaths and serious injuries. The Safety Board believes that authorities At other major airports, who may be having difficulty with their required emergency response times, should follow the example set at Los Angeles International Airport. Slide/Raft Inadequacies Because of the intense fire on the left side of the aircraft, passengers exited from the right side. All cabin exit doors were opened, and all slide/rafts, except the left forwac2d (1L) exit, were deployed. Apparently, the door at exit IL was opened in the emergency mode with the slide deployment mechanism disarmed. When the captain and a male passenger attempted to attach the slide/raft, it was pulled from its door container and fell to the ground. The slide/rafts which were deployed from the four right side emergency exits were exposed to fire and radiant heat. All of t’.e.. slide/rafts failed before the evacuation was completed. Passengers and 10/ Tests have shown that the hydraulic fluid used in aircraft systems, Skydrol, is not fiammable under noraml circumstances. However, when sudjected to heat in a vaporized form under pressure, it will ignite and burn. crewnembers who were still in the aircraft when all slide/rafts had faiied either jumped to the ground or slid down the escape rope from the first officer's side window in the cockpit. The Board understands that the primary purpose of an emergency evacuation system is to provide for rapid pastenger and crew egress from an aircraft under emergency conditions. However, this investigation disclosed that when these slide/rafts were certizicate! as a part of the DC-10 aircraft, no consideration was given to the slide/rafts serviceability when exposed to radiant heat. --- Footnotes: [10/ Tests have shown that the hydraulic fluid used in aircraft systems, Skydrol, is not fiammable under noraml circumstances. However, when sudjected to heat in a vaporized form under pressure, it will ignite and burn.]

[ANALYSIS] However, as previously stated, the Board believes that these events occurred because the flight crew did not use the available cues on the airport surface during taxi and did not cross-check and confirm the airplane’s position on the runway before departure. Also, the flight crewmembers engaged in a nonpertinent conversation during a critical phase of flight (taxi operations), which caused them to lose positional awareness. In addition, before the airplane arrived at the dark runway, the first officer briefed the outage of the runway end identifier lights, he recounted that lights were out during his arrival at LEX early on the morning that preceded the accident flight, and the flight crew most likely read the NOTAM in the flight release paperwork that indicated that the runway 4/22 centerline lights were out of service. This accident is not the first one involving a wrong runway takeoff in which pilots did not use existing cues to identify the airplane’s location on the airport surface or cross-check and verify the airplane’s position before takeoff. For example, the final report on the October 31, 2000, Singapore Airlines flight 006 accident (see section 1.18.4) stated that the pilots did not verify the airplane’s position on the taxi route as the airplane was turning onto the wrong runway and concluded that the flight crew lost situational awareness and took off from the wrong runway despite numerous available cues that provided information about the airplane’s position on the airport. However, the final report for that accident cited a possible reason for the flight crew’s actions: the pilots did not adequately review the taxi route to ensure that they understood that the route to the correct departure runway required passing a parallel runway that was under construction and was open only for taxi operations. Numerous other wrong runway events have occurred. The ASRS database showed 114 reports of incidents from March 1988 to September 2005 in which flight crews of turbojet airplanes lined up on the wrong runway for takeoff. The ASRS reports indicated that the pilots involved in these events, pilots of other aircraft in the area at the time, or air traffic controllers detected the mistake either before or after takeoff. Also, on October 30, 2006, an Alaska Airlines 737 departed from the wrong runway at SEA. According to postincident interviews, the controller instructed the flight crew to taxi the airplane into position and hold on runway 34C, but the captain assumed that the airplane would be taking off from runway 34R. After the airplane departed uneventfully from runway 34R, the controller informed the flight crew that the airplane had departed from the wrong runway. Most recently, on April 18, 2007, a United Airlines Airbus A320 taxied onto runway 27, which was closed, instead of runway 30, the assigned departure runway, at MIA and began the takeoff roll. A NOTAM, which was included in the flight release paperwork, and the ATIS information broadcast, which the flight crew received, indicated Analysis National Transportation Safety Board A I R C R A F T Accident Report 76 that runway 27 was closed.

[FINDINGS] 103 Conclusions 3. Findings 3.1  The captain and the first officer were properly certificated and qualified under Federal 1. regulations. There was no evidence of any medical or behavioral conditions that might have adversely affected their performance during the accident flight. Before reporting for the accident flight, the flight crewmembers had rest periods that were longer than those required by Federal regulations and company policy. The accident airplane was properly certified, equipped, and maintained in accordance 2. with Federal regulations. The recovered components showed no evidence of any structural, engine, or system failures. Weather was not a factor in this accident. No restrictions to visibility occurred during 3. the airplane’s taxi to the runway and the attempted takeoff. The taxi and the attempted takeoff occurred about 1 hour before sunrise during night visual meteorological conditions and with no illumination from the moon. The captain and the first officer believed that the airplane was on runway 22 when 4. they taxied onto runway 26 and initiated the takeoff roll. The flight crew recognized that something was wrong with the takeoff beyond the 5. point from which the airplane could be stopped on the remaining available runway. Because the accident airplane had taxied onto and taken off from runway 26 without 6. a clearance to do so, this accident was a runway incursion. Adequate cues existed on the airport surface and available resources were present in 7. the cockpit to allow the flight crew to successfully navigate from the air carrier ramp to the runway 22 threshold. The flight crewmembers’ nonpertinent conversation during the taxi, which was not in 8. compliance with Federal regulations and company policy, likely contributed to their loss of positional awareness. The flight crewmembers failed to recognize that they were initiating a takeoff on the 9. wrong runway because they did not cross-check and confirm the airplane’s position on the runway before takeoff and they were likely influenced by confirmation bias. Even though the flight crewmembers made some errors during their preflight activities 10. and the taxi to the runway, there was insufficient evidence to determine whether fatigue affected their performance. Conclusions National Transportation Safety Board A I R C R A F T Accident Report 104 The flight crew’s noncompliance with standard operating procedures, including the 11. captain’s abbreviated taxi briefing and both pilots’ nonpertinent conversation, most likely created an atmosphere in the cockpit that enabled the crew’s errors. The controller did not notice that the flight crew had stopped the airplane short of the 12. wrong runway because he did not anticipate any problems with the airplane’s taxi to the correct runway and thus was paying more attention to his radar responsibilities than his tower responsibilities. The controller did not detect the flight crew’s attempt to take off on the wrong runway 13. because, instead of monitoring the airplane’s departure, he performed a lower-priority administrative task that could have waited until he transferred responsibility for the airplane to the next air traffic control facility.

[ANALYSIS] If the controller had noticed that the airplane had stopped in that location before issuing the takeoff clearance, the controller could have queried the flight crew, issued additional taxi instructions, or closely monitored the airplane’s subsequent progress. The controller’s postaccident statements indicated that he did not notice that the airplane had stopped short of runway 26. As a result, he missed an opportunity to prevent the flight crew’s surface navigation error and subsequent wrong runway takeoff attempt. The second time period lasted 28 seconds. It began about 0605:56, when the airplane began to align with runway 26, and ended about 0606:24, when the airplane accelerated beyond the maximum airspeed that would have allowed the airplane to remain on the available runway if the flight crew rejected the takeoff and used maximum braking. The airplane’s movements during this time were not consistent with the clearance provided by the controller and were a clear sign of a lack of positional awareness on the part of the flight crew. If the controller had been looking out the tower cab window and monitoring the flight, he could have addressed this situation by alerting the flight crewmembers that the airplane was on the wrong runway or, later, by instructing them to reject the takeoff. The controller did not take any such actions. The controller indicated that he did not see the airplane align with runway 26 or begin its takeoff roll because he had turned around to perform the traffic count, which is an administrative record-keeping task. The Safety Board examined possible reasons why the controller did not notice indications of the flight crew’s surface navigation error either during the 50-second window of opportunity, which began about 2 minutes before the accident, or during the 28-second critical window, which began about 39 seconds before the accident. Figure 7 shows a detailed timeline of communications and events that occurred in the 2 minutes before the accident. 192  Flight crews stop along a taxi route for various reasons, including a perception of potential traffic conflicts, passenger movements in the cabin, and uncertainty about the taxi clearance or the taxi routes. Some stops during taxi are accompanied by radio transmissions to ATC to explain the delay or seek clarification. Some controllers have a heightened sense of awareness when stops are made during taxi depending on the circumstances and the duration. Analysis National Transportation Safety Board A I R C R A F T Accident Report 78 Air Traffic Control Event Timeline Figure 7.  Note: COM, Comair; EGF, American Eagle; SKW, SkyWest; ZID, Indianapolis Air Route Traffic Control Center; TMU, traffic management unit. Analysis National Transportation Safety Board A I R C R A F T Accident Report 79 Window of Opportunity During Which the Airplane Was Stopped 2.2.3.1  Short of the Wrong Runway On the day of the accident, all LEX tower and radar positions were combined and were being worked by one controller.193 As a result, the controller had to switch his attention between tower and radar tasks.

[ANALYSIS] At this position, the flight crew would have been able to see the runway 26 holding position sign, the “26” painted runway number, the taxiway A lights across runway 26, and the runway 22 holding position sign in the distance.151 FDR data showed that, about 0605:24, the captain began to taxi the airplane across the runway 26 hold short line. FDR data also showed that, about 0605:41, the airplane began to turn onto runway 26, and the CVR showed that, about 0605:46, the first officer completed the lineup checklist. 150  In accordance with FAA Order 7110.65, “Air Traffic Control,” paragraph 3-7-2, the controller would have been required to provide turn-by-turn directions to the departure runway if the flight crew had so requested. 151  During the 50-second timeframe, the controller did not query the flight crew regarding why the airplane was stopped at the hold short line for runway 26. The controller’s actions during the taxi and attempted takeoff sequence are discussed in section 2.2.3. Analysis National Transportation Safety Board A I R C R A F T Accident Report 60 Takeoff Roll 2.2.1.3  About 0605:58, the captain transferred control of the airplane to the first officer, stating, “all yours,” to which the first officer acknowledged, “my brakes, my controls.” At this time, the captain would have switched his attention from outside to inside the cockpit, and the first officer would have switched his attention from inside to outside the cockpit. About 2 seconds later, the airplane was aligned with the centerline for runway 26. The CVR recording showed that the flight crew had referred to runway 22 as the departure runway multiple times before takeoff, and FDR data showed that the pilots’ heading bugs were set to 227º, which was consistent with the magnetic heading for runway 22. The Safety Board concludes that the captain and the first officer believed that the airplane was on runway 22 when they taxied onto runway 26 and initiated the takeoff roll. About 0606:16, the first officer stated, “[that] is weird with no lights,” to which the captain responded, “yeah.” At that time, the airplane was passing through the intersection of runway 26 with runway 22. About 0606:24, the captain called the 100‑knot airspeed check. About the same time, the airplane accelerated beyond the maximum airspeed that would have allowed the airplane to remain on the available runway if the flight crew had rejected the takeoff and used maximum braking.152 At 0606:31.2, the captain called, “V one, rotate,” followed immediately by his exclamation, “whoa.” The aircraft performance study for this accident showed that, at the time of the VR callout, the airplane was 236 feet from the end of the runway.

[ANALYSIS] Therefore, the Safety Board reiterates Safety Recommendations A-00-67 and -68. In addition, no FAA guidance specifically prohibits issuing a takeoff clearance until all intersecting runways to the departure runway have been crossed. On January 4, 2007, the LEX air traffic manager issued a notice that stated that controllers at the tower were not to issue takeoff clearances for runway 22 until the departing airplanes were observed to have completely crossed runway 26.220 Such guidance would benefit other airports with intersecting runways. On June 1, 2007, the FAA issued Notice N JO 7110.468 to amend the required phraseology for issuing departure instructions. According to this notice, a controller has to specifically clear an airplane across all intervening runways before issuing a takeoff clearance. However, this guidance does not instruct controllers to wait until an airplane has crossed the runways before issuing the takeoff clearance. 219  (a) Reports by Airline Pilots on Airport Surface Operations: Part 2. Identified Problems and Proposed Solutions for Surface Operational Procedures and Factors Affecting Pilot Performance, Technical Report No. MTR94W0000060.v2, McLean, Virginia: Mitre Corporation, 1994; and (b) Reports by Air Traffic Control Tower Controllers on Airport Surface Operations: The Causes and Prevention of Runway Incursions, Technical Report No. MTR98W0000033, McLean, Virginia: Mitre Corporation, 1998. 220  In addition, in its July 17, 1989, letter transmitting Safety Recommendation A-89-74 (see section 1.18.3.1), the Safety Board noted that the controllers at HOU were required to observe airplanes cross the approach end of runway 17 before issuing a clearance for takeoff for runway 12. This requirement was the result of two pilot deviation events in early 1989 that involved departures of U.S. air carrier airplanes from the wrong runway at the airport. Analysis National Transportation Safety Board A I R C R A F T Accident Report 97 The Safety Board concludes that, if controllers were required to delay a takeoff clearance until confirming that an airplane has crossed all intersecting runways to a departure runway, the increased monitoring of the flight crew’s surface navigation would reduce the likelihood of wrong runway takeoff events. Therefore, the Safety Board believes that the FAA should prohibit the issuance of a takeoff clearance during an airplane’s taxi to its departure runway until after the airplane has crossed all intersecting runways.

[ANALYSIS] Analysis National Transportation Safety Board A I R C R A F T Accident Report 61 the flight crew had not correctly interpreted these cues or noticed them until after it was too late to successfully abort the takeoff. The Safety Board concludes that the flight crew recognized that something was wrong with the takeoff beyond the point from which the airplane could be stopped on the remaining available runway. Runway Incursions 2.2.1.4  The FAA currently defines a runway incursion as “any occurrence in the airport runway environment involving an aircraft, vehicle, person, or object on the ground that creates a collision hazard or results in a loss of required separation with an aircraft taking off, intending to take off, landing, or intending to land.” ICAO defines a runway incursion as “any occurrence at an aerodrome involving the incorrect presence of an aircraft vehicle or person on the protected area of a surface designated for the landing and take-off of aircraft.” According to these definitions, ICAO would classify this accident as a runway incursion, but the FAA would not consider this accident to be a runway incursion because no other airplane impeded the accident airplane’s ability to take off. The Safety Board notes that the presence of another airplane on a runway should not be a consideration in determining whether a runway incursion has occurred; rather, criteria for making this determination should consider, among other things, whether an airplane’s movement is consistent with the clearances provided to the flight crew. As a result, the Safety Board concludes that, because the accident airplane had taxied onto and taken off from runway 26 without a clearance to do so, this accident was a runway incursion.155 At the Safety Board’s runway incursion forum in March 2007, the FAA announced that it planned to revise its definition of a runway incursion to align with ICAO’s definition by the end of fiscal year 2007 and that it would begin reporting runway incursions according to the revised definition in fiscal year 2008. The Board is encouraged by the FAA’s plan to adopt and apply the ICAO definition because wrong runway takeoffs should be reflected in runway incursion statistics so that runway safety trends can be accurately monitored and appropriate countermeasures can be taken. Pilot Human Factors 2.2.2  The flight crew’s performance on the day of the accident seemed to be uncharacteristic with past reports. The captain and the first officer were described favorably by company personnel, and pilots who had flown with them described both as competent pilots who had not previously demonstrated difficulty with airport surface operations. The captain was described as someone who managed the cockpit well, adhered to standard operating procedures, and demonstrated good CRM. The first officer was preparing for an opportunity to upgrade to captain and was described as someone who would have made a good captain because of his adherence to standard operating procedures. 155  Even though the FAA does not consider the SEA and MIA incidents (discussed in section 1.18.4) to be runway incursions, the Safety Board does consider these incidents to be runway incursions.

[ANALYSIS] The appearance of the runway end environment would have provided a salient cue to the flight crew that the airplane was in an extremely hazardous situation and could not remain on the ground. The airplane’s airspeed at the time of the captain’s VR callout was 131 knots, which was 11 knots below the planned VR airspeed of 142 knots (which the flight crew had briefed and entered into the airplane’s EFIS).153 Thus, the captain’s early VR callout and subsequent “whoa” exclamation indicated that he recognized that something was wrong with the takeoff. FDR data showed that, in response, the first officer pulled the control column full aft154 and that the airplane rotated at a rate of about 10º per second, which is three times the normal rotation rate. This abnormal column input showed that the first officer also recognized that something was wrong with the takeoff. Although numerous cues, including the lack of runway lighting, were available to indicate that the airplane was not on the assigned runway (see sections 2.2.2.1 and 2.2.2.3), 152  According to calculations by Bombardier, for the airplane to have stopped before the end of runway 26, maximum braking would have had to start when the airplane was at an airspeed of about 103 knots. 153  FDR data for the accident airplane’s 12 previous takeoffs indicated that rotation occurred at or after reaching the VR airspeed. 154  FDR data showed that the left and right control column inputs during the accident rotation reached 10.6º and 10.9º, respectively, and that the nominal control column input for rotation during the accident airplane’s 12 previous takeoffs was between 4º and 5º. Analysis National Transportation Safety Board A I R C R A F T Accident Report 61 the flight crew had not correctly interpreted these cues or noticed them until after it was too late to successfully abort the takeoff. The Safety Board concludes that the flight crew recognized that something was wrong with the takeoff beyond the point from which the airplane could be stopped on the remaining available runway.

[ANALYSIS] Analysis National Transportation Safety Board A I R C R A F T Accident Report 59 instruction. Because two airplanes, SkyWest flight 6819 and American Eagle flight 882, were given the same taxi clearance and had already correctly taxied to and held short of runway  22 without any special instructions, there was no apparent reason for the controller to have suspected that the pilots would have had difficulty navigating to the departure runway.150 Taxi to Runway 2.2.1.2  From about 0603:16 to 0603:56, while the captain was taxiing the airplane and performing navigational checking activities, both pilots resumed the nonpertinent discussion that was started while the airplane was parked at the gate. (Figure 1 shows the location of the airplane along the taxi route while this conversation was occurring.) The nonpertinent conversation was not in compliance with the sterile cockpit rule required by company procedures and 14 CFR 121.542 (see section 1.17.1.3). The primary reason for the sterile cockpit rule is to ensure that the pilots’ attention is directed to operational concerns during critical phases of flight (including taxi) and is not redirected or degraded because of nonessential activities or discussion. FDR data showed that, about 0604:33, the airplane stopped on taxiway A at the hold short line for runway 26, which was about 560 feet from the intended destination— the hold short line for runway 22. During this time, the first officer was completing the before takeoff checklist. About 0605:15, the first officer advised the controller that the airplane was ready to depart, and the controller told the flight crew that the airplane should fly the runway heading and was cleared for takeoff. Neither the first officer nor the controller stated the runway number during the request and clearance for takeoff, but ATC procedures did not require them to do so. Because the flight crew believed that the airplane was at the hold short line for runway 22 at the time of the takeoff clearance (see section 2.2.2.3), the absence of a reference to runway 22 in the request and clearance for takeoff was not a factor in this accident. The 50-second timeframe during which the airplane was stopped at the runway 26 hold short line should have provided the flight crew with ample time to look outside the cockpit and determine the airplane’s position on the airport. At this position, the flight crew would have been able to see the runway 26 holding position sign, the “26” painted runway number, the taxiway A lights across runway 26, and the runway 22 holding position sign in the distance.151 FDR data showed that, about 0605:24, the captain began to taxi the airplane across the runway 26 hold short line.

[CONCLUSIONS] The controller did not detect the flight crew’s attempt to take off on the wrong runway 13. because, instead of monitoring the airplane’s departure, he performed a lower-priority administrative task that could have waited until he transferred responsibility for the airplane to the next air traffic control facility. The controller was most likely fatigued at the time of the accident, but the extent 14. that fatigue affected his decision not to monitor the airplane’s departure could not be determined in part because his routine practices did not consistently include the monitoring of takeoffs. The Federal Aviation Administration’s operational policies and procedures at the 15. time of the accident were deficient because they did not promote optimal controller monitoring of aircraft surface operations. The first officer’s survival was directly attributable to the prompt arrival of the first 16. responders; their ability to extricate him from the cockpit wreckage; and his rapid transport to the hospital, where he received immediate treatment. The emergency response for this accident was timely and well coordinated. 17. A standard procedure requiring 14 18. Code of Federal Regulations Part 91K, 121, and 135 pilots to confirm and cross-check that their airplane is positioned at the correct runway before crossing the hold short line and initiating a takeoff would help to improve the pilots’ positional awareness during surface operations. The implementation of cockpit moving map displays or cockpit runway alerting 19. systems on air carrier aircraft would enhance flight safety by providing pilots with improved positional awareness during surface navigation. Enhanced taxiway centerline markings and surface painted holding position 20. signs provide pilots with additional awareness about the runway and taxiway environment. This accident demonstrates that 14 21. Code of Federal Regulations 91.129(i) might result in mistakes that have catastrophic consequences because the regulation allows an airplane to cross a runway during taxi without a pilot request for a specific clearance to do so. Conclusions National Transportation Safety Board A I R C R A F T Accident Report 105 If controllers were required to delay a takeoff clearance until confirming that an 22. airplane has crossed all intersecting runways to a departure runway, the increased monitoring of the flight crew’s surface navigation would reduce the likelihood of wrong runway takeoff events. If controllers were to focus on monitoring tasks instead of administrative tasks when 23. aircraft are in the controller’s area of operations, the additional monitoring would increase the probability of detecting flight crew errors. Even though the air traffic manager’s decision to staff midnight shifts at Blue Grass 24. Airport with one controller was contrary to Federal Aviation Administration verbal guidance indicating that two controllers were needed, it cannot be determined if this decision contributed to the circumstances of this accident. Because of an ongoing construction project at Blue Grass Airport, the taxiway identifiers 25. represented in the airport chart available to the flight crew were inaccurate, and the information contained in a local notice to airmen about the closure of taxiway A was not made available to the crew via automatic terminal information service broadcast or the flight release paperwork.

[ANALYSIS] In addition, the angle from the runway 26 hold short line on taxiway A to runway 26 was the same as the angle from the runway 22 hold short line on former taxiway A (north of runway 8/26) to runway 22. Also, the taxiway A centerline split into three lines after the runway 26 hold short line. These extended taxiway centerlines led onto the closed portion of taxiway A, across runway 26 to runway 22, and onto the runway 26 centerline (a lead-on/off line). The presence of a lead-on/off line from a taxiway directly to a runway could have supported the captain’s perception that the airplane had arrived at the departure runway. In addition, as stated in section 2.2.1.1, the first officer told the captain during his takeoff briefing, “lights are out all over the place,” in reference to observations he made while landing on runway 22 about 0140 on the day before the accident. The first officer’s statement might have contributed to the captain’s perception that the airplane was taxiing onto runway 22 because he might have anticipated a dark runway environment. At the 165  R. Conejo and C.D. Wickens, The Effects of Highlighting Validity and Feature Type on Air‑to‑Ground Target Acquisition Performance, Technical Report ARL-97-11, Savoy, Illinois: University of Illinois Aviation Research Lab, 1997. 166  C.D. Wickens and J.S. McCarley, Attention-Situation Awareness (A-SA) Model of Pilot Error, Technical Report ARL-01-13/NASA-01/06, Savoy, Illinois: University of Illinois Aviation Research Lab, 2001. 167  Jason S. McCarley, Margaret J. Vais, Heather Pringle, Arthur F. Kramer, David E. Irwin, and David L. Strayer, “Conversation Disrupts Change Detection in Complex Traffic Scenes,” Human Factors, vol. 46, no. 3, pages 424-436, 2004. Analysis National Transportation Safety Board A I R C R A F T Accident Report 66 least, this statement reduced the significance of the first officer’s subsequent statement, “[that] is weird with no lights” (to which the captain responded “yeah”), as the airplane rolled down runway 26. There are well-known psychological concepts associated with perception and decision-making that can allow a person’s mistaken assessment to persist. For example, confirmation bias occurs when people seek out or observe elements in their environment that support their perception. Specifically, confirmation bias results from a tendency for people to primarily seek out confirming evidence of a belief while spending less effort to seek out negative evidence that can disconfirm the belief.168 Confirmation bias can cause a person to persist in holding an incorrect belief despite the availability of contradictory evidence.

[ANALYSIS] Analysis National Transportation Safety Board A I R C R A F T Accident Report 67 At the time that the first officer began to increase thrust for takeoff, FDR data showed that the magnetic heading of the airplane was about 266º, which corresponded to the magnetic heading for runway 26. Figures 6 and 6a show the approximate configuration of the captain’s MFD and PFD, including the heading bug setting and magnetic heading indication, when the airplane was lined up on the runway 26 centerline. As shown in the figures, the heading bug was offset by 40º, providing a salient cue that the airplane was not lined up on the correct departure runway. The CVR did not record any awareness by the flight crewmembers about this offset. Multifunction Display Figure 6.  Source: Rockwell Collins. The display was configured as requested by the Safety Board. Analysis National Transportation Safety Board A I R C R A F T Accident Report 68 Primary Flight Display Figure 6a. Source: Rockwell Collins. The display was configured as requested by the Safety Board. The wayfinding task includes an ongoing cross-check between an airplane’s expected and actual position using available cues in the environment and aids in the cockpit. The CVR did not record any discussion by the flight crew about the need to cross-check the airplane’s position on the runway.171 The Safety Board concludes that the flight crewmembers failed to recognize that they were initiating a takeoff on the wrong runway because they did not cross-check and confirm the airplane’s position on the runway before takeoff and they were likely influenced by confirmation bias. 171  On March 23, 2007, Comair revised its operations manual to include a departure runway checklist item. Analysis National Transportation Safety Board A I R C R A F T Accident Report 69 Fatigue 2.2.2.4  The Safety Board examined whether pilot fatigue could have been a factor in this accident by assessing the preconditions that could allow for the development of a fatigued state and examining the nature of the pilots’ demonstrated performance deficiencies. Potential conditions that can lead to the development of a fatigued state include chronic sleep restriction, acute sleep loss, circadian disruption (work during times when one would normally be asleep), and time since awakening. The captain and the first officer received more than the minimum required rest periods during their respective trips in the days before the accident, and their flight and duty times in the week and month before the accident would not have precluded them from obtaining adequate sleep.

[ANALYSIS] Preflight Activities and Actions During the Taxi 2.2.2.2  Because the availability of cues and aids for the pilots’ wayfinding task was not a factor in this accident, the Safety Board examined the crew’s actions during the preflight and taxi phases of the flight’s operation to identify possible reasons for the error. The flight crew proceeded from the Comair operations center to the air carrier ramp area, where two Comair CRJ airplanes were located. The crew initially boarded the wrong airplane,162 even though the tail number of the airplane to be used for the flight 158  Such communications would have been consistent with good CRM (while at towered airports) and company policy. In addition, this behavior would have been consistent with reports of the pilots’ past performance. 159  The SkyWest first officer stated that he became “momentarily confused” about the airplane’s orientation when he looked up while crossing runway 26 but that he was able to reorient himself after identifying the holding position sign for runway 22. 160  The SkyWest and American Eagle pilots did not recall seeing the low-profile barriers on taxiway A north of runway 26. The pilots’ lack of recall in this area indicates that the barriers were not significant elements in the pilots’ wayfinding and that the taxiway A closure did not encumber their navigation to runway 22. The pilots’ lack of recall does not indicate that the barriers were inconspicuous or difficult to see when approaching or crossing runway 26. In fact, observations in a CRJ-100 airplane after the accident indicated that the barriers would have been clearly visible to the pilots during the taxi. The pilots’ failure to recall these barriers is not unusual because people can have difficulty remembering things in their environment with which they did not interact. 161  After runway 26 was reopened (November 2006), two events occurred in which air traffic controllers informed pilots that their airplanes were on that runway instead of runway 22 (see section 1.10.4). However, an interview with the pilot involved in the first event indicated that he had stopped for preflight activities, was aware of the airplane’s position, and did not intend to depart on runway 26. An interview with the captain of the flight involved in the second event indicated that he became temporarily confused about the location of runway 22 but that the airplane was never lined up to take off from runway 26. An interview with the first officer of this flight indicated that he and the captain were never confused about the correct runway for takeoff. 162  The Safety Board does not know which pilot was the first one to board the wrong airplane. Analysis National Transportation Safety Board A I R C R A F T Accident Report 64 was included in the flight release paperwork, and started its APU. Although these actions likely consumed a portion of the crew’s available time at the gate, CVR evidence and interviews with ground personnel indicated that neither pilot appeared to be rushed or hurried as he completed required tasks.

[ANALYSIS] Even though discrepancies existed between the airport chart and the external cues available to the pilots because of an ongoing construction project at the airport,157 the chart depicted the paved taxiway and runway surfaces at the time of the accident. Another available resource within the cockpit was the instrumentation, including the heading bug, which had been set to 227º to correspond to the magnetic heading for runway 22. This heading information, which was clearly presented on both flight crewmembers’ MFDs and PFDs, would have provided the pilots with a real-time cue of their orientation relative to runway 22. 156  Although the first officer’s attention was focused inside the cockpit while the airplane was taxiing to the departure runway, he would have had multiple opportunities to look outside the cockpit and monitor the airplane’s progress along the taxi route. Such monitoring would have helped the first officer gauge and pace his activities with the available time left for taxi. 157  Taxiway A north of runway 26 had been closed (as indicated by a local NOTAM) and barricaded, and taxiway A5 had been redesignated as taxiway A. Neither of these changes was depicted on the airport chart. However, no evidence indicated that either pilot was confused by the discrepancies. In addition, the pilots of SkyWest flight 6819 and American Eagle flight 882 used the same chart to navigate to runway 22. Analysis National Transportation Safety Board A I R C R A F T Accident Report 63 In addition, the flight crewmembers could have communicated with the controller if they became unsure of their position at any time.158 However, the CVR did not record any statement by either flight crewmember about being unsure of the airplane’s position on the airport surface at any time during the taxi and takeoff roll. The CVR also did not record any indication that the crewmembers had attempted to confirm the airplane’s position on the runway before beginning the takeoff roll. The taxi routing and cues available to the accident pilots were identical to those that were available to the pilots of the two regional jets (SkyWest flight 6819 and American Eagle flight 882) that departed before Comair flight 5191. The SkyWest and American Eagle pilots had no difficulty identifying, and successfully navigating to, runway 22 using the available cues,159 even with the differences in taxiway signage and chart labeling.160 In addition, even though the airport configuration at the time of the accident had been in place for 1 week, the Safety Board did not identify any reports about surface navigation problems at LEX during that time.161 The Safety Board concludes that adequate cues existed on the airport surface and available resources were present in the cockpit to allow the flight crew to successfully navigate from the air carrier ramp to the runway 22 threshold. Preflight Activities and Actions During the Taxi 2.2.2.2  Because the availability of cues and aids for the pilots’ wayfinding task was not a factor in this accident, the Safety Board examined the crew’s actions during the preflight and taxi phases of the flight’s operation to identify possible reasons for the error.

[ANALYSIS] Because no other traffic was on the airport surface to pose a conflict to the airplane during its taxi, the controller would not have expected much useful information to be obtained by frequently scanning the runway environment, which would have decreased the likelihood that he was frequently looking out the tower cab windows at the runway environment during the 50-second window of opportunity. The Safety Board concludes that the controller did not notice that the flight crew had stopped the airplane short of the wrong runway because he did not anticipate any problems with the airplane’s taxi to the correct runway and thus was paying more attention to his radar responsibilities than his tower responsibilities. Postaccident observations from the LEX ATCT revealed that, from the controller’s work station at night, it was somewhat difficult to see whether the CRJ-100 demonstration airplane was located at the hold short line for runway 26, on taxiway A, or at the hold short line for runway 22 because of (1) the proximity of these locations in the controller’s visual field as a result of the new taxiway configuration198 and (2) the reduced visibility of ground texture, linear perspective, and other monocular depth cues that are useful for judging distances beyond 15 to 20 feet. Although the controller had been working at the tower for 17 years and had presumably become an expert at recognizing aircraft positions on the airport surface, the use of former taxiway A5 (redesignated as taxiway A) to reach the runway 22 threshold was new to him because the runway had been shifted and the redesignated taxiway had been in place for only 1 week (four of the controller’s shifts) 198  From the tower cab, the hold short lines for runways 22 and 26 appeared close to each other because they were separated by less than 5º of visual angle. Analysis National Transportation Safety Board A I R C R A F T Accident Report 81 at the time of the accident.199 These factors would have made it more difficult for the controller to determine the airplane’s exact location. In addition, the controller was not required to determine that the airplane had reached the departure runway before he cleared the airplane for takeoff; he was only expected to determine that the airplane was at a location that was consistent with its taxi clearance. When the flight crew had stopped the airplane at the runway 26 hold short line, the airplane was in a location that was consistent with its taxi clearance. The controller reported that he did not see the airplane stop in this position. Even if the controller had seen the airplane at that time and noticed that it was not moving, a brief scan of the runway environment would not have informed him of whether the airplane had been stopped only briefly or for a longer period of time. Nevertheless, the controller could have detected that the airplane had stopped short of the wrong runway if he had been monitoring the airplane’s progress along the taxi route.

[ANALYSIS] LEX is not one of those airports; during 2005, it had about 0.5 million passenger enplanements. 218  AC 150/5340-1J also stated, “installation at other airports is at the option of the airport operator. If an airport operator decides to exercise this option, the enhanced markings must be installed at all holding positions on the airport.” Analysis National Transportation Safety Board A I R C R A F T Accident Report 96 Taxi and Takeoff Clearances 2.4.4  As stated in section 2.2.1.1, 14 CFR 91.129(i) permits pilots, after receiving taxi clearance, to cross all intersecting runways along the taxi route (without stopping) except for the assigned departure runway. On July 6, 2000, the Safety Board issued Safety Recommendations A-00-67 and -68, which asked, in part, that the FAA (1) amend 14 CFR 91.129(i) to require that all runway crossings be authorized only by specific ATC clearance and (2) amend FAA Order 7110.65 to require that, for aircraft that need to cross multiple runways, air traffic controllers issue an explicit crossing instruction for each runway after the previous runway has been crossed. The Board classified these safety recommendations “Open—Unacceptable Response” on April 11, 2006. If these safety recommendations had been implemented before this accident, the controller would have been required to issue a specific taxi clearance for the airplane to cross runway 26 and then issue a specific taxi clearance for the airplane to continue taxiing to runway 22. These procedures would have provided the flight crew with better awareness of the airplane’s position along the taxi route and would have required the controller to visually observe the airplane’s position and monitor the taxi as the airplane progressed toward the departure runway. Thus, the flight crew’s surface navigation error might have been prevented. In addition, Mitre reports cited pilot and controller concerns about the adequacy of runway crossing requirements, and most of these pilots and controllers thought that it would be beneficial to safety to modify 14 CFR 91.129(i) so that it required a specific clearance for each runway crossing.219 The Safety Board concludes that this accident demonstrates that 14 CFR 91.129(i) might result in mistakes that have catastrophic consequences because the regulation allows an airplane to cross a runway during taxi without a pilot request for a specific clearance to do so. Therefore, the Safety Board reiterates Safety Recommendations A-00-67 and -68. In addition, no FAA guidance specifically prohibits issuing a takeoff clearance until all intersecting runways to the departure runway have been crossed.

[ANALYSIS] These interventions can also help prevent runway incursions, which is an issue on the Safety Board’s list of Most Wanted Transportation Safety Improvements. 210  This definition was cited in the Safety Board’s 1981 special study on cabin safety in large transport aircraft. According to the study, the definition (1) was developed using aviation crash injury research by Cornell University and aviation safety engineering and research by the Flight Safety Foundation and (2) was used in the Aircraft Crash Survival Design Guide, which was prepared by the U.S. Army Research and Technology Laboratories along with other Federal agencies. The definition has been used by the Board since that time to determine the survivability of accidents. In addition, the definition appeared in the Board’s 2001 safety report on the survivability of accidents involving Part 121 U.S. air carrier operations from 1983 to 2000. Analysis National Transportation Safety Board A I R C R A F T Accident Report 91 Flight Deck Procedures 2.4.1  Well-designed flight deck procedures can be an effective countermeasure against surface operation errors. After this accident, the Safety Board recognized the need to improve industry standards for confirming an airplane’s position at the departure runway before takeoff and, on December 12, 2006, issued Safety Recommendation A-06‑83. This recommendation asked that the FAA require all Part 121 operators to establish procedures requiring all crewmembers on the flight deck to positively confirm and cross-check the airplane’s location at the assigned departure runway before crossing the hold short line for takeoff. On April 16, 2007, the FAA issued SAFO 07003, “Confirming the Takeoff Runway,” in response to Safety Recommendation A-06-83. According to the SAFO, its purpose is to emphasize the importance of implementing standard operating procedures and training for flight crews to ensure that an airplane is at the intended runway. SAFO 07003 was aimed at directors of safety, directors of operations, fractional ownership program managers, trainers, and pilots. The SAFO stated that pilots should positively confirm and cross-check the takeoff runway and the airplane’s location at the assigned departure runway before crossing the hold short line and while in the takeoff position. The SAFO further stated that airplane-specific standard operating procedures should be established, implemented, and supported by pilot training that uses all available resources to confirm and cross-check an airplane’s position. The SAFO mentioned that these resources included the HSIs, which can confirm that an airplane’s position is where the flight crew intended, and air traffic controllers, who can help confirm an airplane’s position during taxi or at a hold short line. The Safety Board is encouraged that the FAA is providing renewed emphasis about the importance of cross-checking and confirming an airplane’s position on a runway.

[ANALYSIS] For example, confirmation bias occurs when people seek out or observe elements in their environment that support their perception. Specifically, confirmation bias results from a tendency for people to primarily seek out confirming evidence of a belief while spending less effort to seek out negative evidence that can disconfirm the belief.168 Confirmation bias can cause a person to persist in holding an incorrect belief despite the availability of contradictory evidence. For the flight crew, confirmation bias was in place not only at the hold short line for runway 26 but also during the initial acceleration down the runway because the crew did not evaluate evidence that would contradict the airplane’s position on the airport surface at the time. There were cues available to the flight crew that were not consistent with a taxi onto runway 22. These cues included the runway holding position sign for runway 26, the 75-foot painted width of runway 26 (versus the 150-foot width of runway 22),169 and the absence of runway edge lights and precision runway markings (such as threshold markings and touchdown zone markings) on runway 26. However, once the flight crewmembers had navigated to what they believed was the correct runway, they were likely no longer giving strong weight to contradictory information as a result of confirmation bias. On July 17, 1989, the Safety Board issued Safety Recommendation A-89-74, which asked the FAA to ensure that the operations manuals of all Part 121 and 135 air carriers require flight crews to cross-check the heading indicator with the runway heading when the airplane is aligned with the runway for takeoff. The Board classified Safety Recommendation A-89-74 “Closed—Acceptable Action” on December 11, 1990, after the FAA revised ACOB 8-85-1 to address the need for pilots to cross-check the heading indicator with the runway heading before takeoff. These bulletins, however, are not mandatory, and, during this investigation, the Board determined that Comair and other Part 121 operators did not have procedures for positively verifying that an airplane was aligned on the correct departure runway.170 FDR data showed that, at some point before the start of the recording, the pilots’ heading bugs had been set to 227º to correspond to the magnetic heading for runway 22. 168  The Safety Board notes that confirmation bias has multiple definitions. This report refers to confirmation bias as a phenomenon that occurs automatically (that is, without conscious intervention or intent) at the perceptual level. 169  Postaccident observations from a CRJ-100 airplane indicated that the reduced available width of runway 26 was clearly visible at night because the side stripes were brightly illuminated by the airplane’s external lighting system. 170  As a result of this finding and the finding that that some Part 121 operators (including Comair) did not provide guidance to their pilots about conducting takeoffs at night on unlighted runways, the Safety Board issued Safety Recommendations A-06-83 and -84 on December 12, 2006 (see sections 1.18.3.1 and 2.4.1).

[ANALYSIS] None of the tires exhibited flat spots or evidence of reverted rubber, and all of the tire tread depths were within specified limits. Therefore, the investigation ruled out the occurrence of reverted rubber hydroplaning. The NASA friction expert reported that the runway was generally in “excellent” condition because it had a relatively high macrotexture, consistently adequate cross-slope throughout the runway length, insignificant rubber deposits, and no concrete surface deterioration. Further, pavement drainage models indicated that the runway was fully capable of draining the rainfall reported throughout the morning of the accident; therefore, there was not sufficient standing water on the runway at the time of the accident to have caused the airplane to experience dynamic hydroplaning. The NTSB concludes that no evidence exists that reverted rubber or dynamic hydroplaning occurred. 2.2.3 Captain’s Decision and Subsequent Attempt to Go Around The pilots remained silent for 10.5 seconds after the captain stated, “no we’re not” (acknowledging that the airbrake handle was not in the DUMP position before moving it to that position), until the captain called out “flaps” at 0945:22. About the same time (more than 17 seconds after touchdown), the CVR recorded a sound consistent with increasing engine noise and the initiation of a go-around. The results of the airplane performance study indicated that, at the time that the go-around was initiated, the deceleration rate was such that the airplane would have exited the runway end at a ground speed of between 23 and 37 knots and stopped between 112 According to the pilot/controller glossary in the AIM, the touchdown zone is “the first 3,000 feet of the runway beginning at the threshold.” NTSB Aircraft Accident Report 62 100 and 300 feet into the 1,000-foot-long runway safety area. Therefore, it can be reasonably assumed that, at some point during the landing roll, the captain likely became concerned that the airplane would run off the runway end and had to decide whether it was preferable to overrun the runway or attempt a go-around. However, as discussed, no evidence indicates that the captain was prepared for the possibility of a go-around. Specifically, he did not conduct an approach briefing, which would have included briefing a missed approach. It is possible that the captain’s decision to go around was delayed because it took time for him to realize that the airplane was not decelerating as he expected and the possibility of a runway overrun was increasing. In addition, he might have been waiting, expecting the airplane’s deceleration to improve.

[ANALYSIS] The flaps were found in the wreckage set to the fully retracted position (flaps 0°), which was an incorrect setting for a go-around and would have made it more difficult for the airplane to lift off, and this setting may reflect the confusion and lack of crew coordination that can follow from unprofessional compliance with use of nonstandard terminology. NTSB Aircraft Accident Report 63 landing roll. The NTSB notes that other recent overrun accidents have not been as catastrophic because the flight crews did not attempt to go around after landing.114 The NTSB has previously investigated accidents during which the pilots did not commit to the landings and made a delayed decision to go around. For example, on October 5, 2005, a Beechcraft 58 overran the runway in Jacksonville, Florida, after attempting a go-around late in the landing roll on a wet, ungrooved runway. During postaccident interviews, the pilot stated that the airplane touched down on the first quarter of the runway at about 100 knots. He stated that, past the midfield point, “the airplane still had a lot more momentum to bleed off,” so, with only one quarter of the runway left, he attempted a go-around. He stated that, when he noticed that the airplane was not climbing, he aborted the go-around and overran the departure end of the runway. In addition, on July 15, 2005, a Cessna 525A collided with a localizer antenna in Newnan, Georgia, after the pilot conducted a go-around late in the landing roll on a wet, ungrooved runway. The pilot stated that he applied brakes upon landing and that the airplane then hydroplaned. He stated that he chose to abort the landing with 2,300 feet of runway remaining (the runway was 5,500 feet long). As a result of the pilot’s delayed decision to go around, the airplane became airborne only 300 feet from the runway end. Both of these accidents might have been prevented if the pilots had committed to the landings or better understood where the committed-to-stop point was rather than attempting to go around with insufficient runway available to lift off and clear obstacles.115 The NTSB concludes that establishing a committed-to-stop point in the landing sequence beyond which a go-around should not be attempted for turbine-powered aircraft would eliminate ambiguity for pilots making decisions during time-critical events.

[ANALYSIS] Specifically, he did not conduct an approach briefing, which would have included briefing a missed approach. It is possible that the captain’s decision to go around was delayed because it took time for him to realize that the airplane was not decelerating as he expected and the possibility of a runway overrun was increasing. In addition, he might have been waiting, expecting the airplane’s deceleration to improve. His expectations for the airplane’s deceleration may have been unrealistic because he may have confused the accident airplane’s performance with that of other company airplanes that were equipped with thrust reversers and previously flown by the captain, or he did not properly account for the tailwind and consequent higher ground speed at touchdown because he was unaware of the changing wind conditions during the approach and landing. According to the East Coast Jets Go-Around checklist, the correct command to set the flaps for a go-around is “flaps 15.”113 Postaccident examination of the flaps indicated that they were fully retracted (that is, set at 0°), not set to 15° as required for a go-around. Takeoff calculations performed by Hawker Beechcraft indicated that, given the location of the initiation of the go-around, the airplane could not have rotated and lifted off before the runway end with the flaps set at either 15° or 0°. Further, ground scars in the grass beyond the runway end (within the runway safety area) indicated that the airplane lifted off about 978 feet beyond the threshold and subsequently impacted the ILS localizer antenna about 5 feet agl. The airplane then continued to roll to the right and descend, impacting the ground. The NTSB concludes that, if the captain had continued the landing and accepted the possibility of overrunning the runway instead of attempting to execute a go-around late in the landing roll, the accident most likely would have been prevented or the severity reduced because the airplane would have come to rest within the runway safety area. A review of the manufacturer, East Coast Jets, and Simcom guidance found no procedures on how to execute a go-around after landing. Further, East Coast Jet pilots were not trained to execute a go-around after landing. However, none of the guidance explicitly states that a go-around should only be conducted before landing or identifies a committed-to-stop point (that is, a point in the landing sequence beyond which a go-around should not be attempted). The captain’s decision to go around more than 17 seconds after touchdown left insufficient runway available to configure the airplane and accelerate to become airborne before reaching the runway end and clearing obstacles. If the captain had conducted an approach briefing that included a committed-to-stop point (for example, in the case of the Hawker Beechcraft 125-800A airplane, once lift dump has been deployed), he may not have decided to attempt a go-around late in the 113 Despite the captain’s use of incorrect terminology, it is clear that his statement, “flaps,” at 0945:22 was part of a go-around attempt and led to a required increase in engine thrust.

[ANALYSIS] The NTSB notes that because the runway was ungrooved and wet, the friction it could provide would have been lower than that achievable on a dry runway, and it would not have required maximum braking effort to achieve the maximum available braking forces on the tires. An increase in brake pressure beyond that required to achieve maximum available braking forces would only tend to lock the wheels, and the brake anti-skid system would then release brake pressure to avoid this situation. The CVR recorded the sound of the airplane touching down at 0945:04 and, at 0945:07, a sound consistent with the airbrake handle moving to the OPEN position. One second later, the first officer stated, “we’re dumped,” and, immediately after, “we’re not dumped.” In response to the first officer’s last statement, the captain stated, “no, we’re not,” while simultaneously making straining sounds, consistent with physically attempting to move a cockpit control. The CVR then recorded a sound consistent with the airbrake handle moving into the DUMP position. According to the airplane performance study, the airplane’s airspeed over the threshold was the reference 60 NTSB Aircraft Accident Report 61 landing airspeed of about 122 knots, and the airplane touched down about 1,128 feet from the runway threshold, which is within the target touchdown zone.112 East Coast Jets procedures call for the immediate deployment of full lift dump upon touchdown. However, the CVR evidence indicates that upon touchdown the captain only moved the airbrake handle to the OPEN position instead of fully aft to the DUMP position and likely did not fully deploy the lift-dump system (full flaps and airbrake deflection) until about 7 seconds after touchdown, which was not in accordance with company procedures (or the deceleration device deployment times used to develop the BAe 125-800A AFM wet runway guidance material). ( See section 2.6.) The captain should have deployed lift dump by moving the airbrake handle in one motion to the DUMP position, not partially deploying the airbrakes and then fully deploying lift dump. The first officer most likely stated, “we’re dumped,” as an automatic callout upon landing when he saw the captain move the airbrake handle aft. The latter callout, “we’re not dumped,” likely resulted from the first officer’s required check of the flap position indicator and provides an example of effective monitoring by the first officer. The NTSB concludes that the airplane touched down within the target touchdown zone and at the recommended touchdown speed and that the captain likely applied sufficient pressure on the brakes during the initial part of the landing roll to take full advantage of the available runway friction, but he failed to immediately deploy the lift-dump system after touchdown in accordance with company procedures, which negatively affected the airplane’s deceleration. None of the tires exhibited flat spots or evidence of reverted rubber, and all of the tire tread depths were within specified limits. Therefore, the investigation ruled out the occurrence of reverted rubber hydroplaning.

[ANALYSIS] The Section 25.109 calculations are consistent with current knowledge about wet runway braking performance, which is reflected in the engineering data used by the FAA to update regulations governing the calculation of accelerate-stop distances for wet, ungrooved runways138 and by the TALPA ARC in drafting new recommendations to require and support arrival landing distance assessments. The NTSB concludes that the wet runway landing distances provided in AFMs or performance supplemental materials that are based on the braking coefficients defined by the BCAR RWHS or AMJ 25X1591 can be significantly shorter than the actual distances required to stop on some wet, ungrooved runways. The NTSB notes that the FAA is expected to initiate rulemaking in 2011 and that, even if the rulemaking is begun this year, it will likely be several years until final adoption and implementation. The NTSB further notes that, although the investigative findings indicated that the airplane would have overrun the runway but remained within the runway safety area if the captain had continued with the landing roll rather than attempted to go around, not all airports with ungrooved runways have safety areas. Therefore, until the FAA has completed the rulemaking, the NTSB recommends that the FAA inform operators of airplanes that have wet runway landing distance data based on the BCAR RWHS or AMJ 25X1591 that the data contained in the AFMs (and/or performance supplemental materials) 137 The NTSB notes that the calculations made in the study assumed that the pilots applied sufficient braking effort to demand all of the available runway friction. If the pilots did not apply sufficient braking effort to demand all of the available runway friction, the landing distances would have been longer than the computed landing distances. 138 The regulations governing the calculation of accelerate-stop distances also provide airframe manufacturers (and ultimately operators) with an option to account for wet, grooved or porous friction course, runway accelerate-stop performance. However, to take this improved wet runway stopping performance credit, a given operator must assume the burden of ensuring (via periodic inspections) that the runway is adequately maintained. NTSB Aircraft Accident Report may underestimate the landing distance required to land on wet, ungrooved runways and work with industry to provide guidance to these operators on how to conduct landing distance assessments when landing on such runways. 2.7 Line Checks According to 14 CFR 135.299, PICs operating under Part 135 must complete an annual line check. The annual line check must be given by an FAA-approved check pilot and consist of at least one flight over one route segment; include takeoffs and landings at one or more representative airports (that is, an airport that the pilot typically flies into); and be flown over a civil airway, an approved off-airway route, or a portion of either.

[CONCLUSIONS > FINDINGS] The NTSB concludes that a lightweight recording system conforming to EUROCAE ED-155 would have helped determine the flight crew’s actions during the landing and subsequent go-around attempt, including, but not limited to, whether they silently conducted checklists (partially or completely), which flap settings they selected, and how much braking effort they made upon landing. 86 NTSB Aircraft Accident Report 3. Conclusions 3.1 Findings 1. The investigation found that the pilots were properly certificated and qualified under federal regulations. 2. The investigation found that the accident airplane was properly certificated, equipped, and maintained in accordance with federal regulations. Examinations of the recovered components revealed no evidence of any preimpact structural, engine, or system failures. The airplane was within normal weight and balance limitations. 3. The accident was not survivable. 4. The captain allowed an atmosphere in the cockpit that did not comply with well-designed procedures intended to minimize operational errors, including sterile cockpit adherence, and this atmosphere permitted inadequate briefing of the approach and monitoring of the current weather conditions, including the wind information on the cockpit instruments; inappropriate conversation; nonstandard terminology; and a lack of checklist discipline throughout the descent and approach phases of the flight. 5. The flight crewmembers exhibited poor aeronautical decision-making and managed their resources poorly, which prevented them from recognizing and fully evaluating alternatives to landing on a wet runway in changing weather conditions, eroded the safety margins provided by the checklists, and degraded the pilots’ attention, thus increasing the risk of an accident. 6. The airplane touched down within the target touchdown zone and at the recommended touchdown speed, and the captain likely applied sufficient pressure on the brakes during the initial part of the landing roll to take full advantage of the available runway friction, but he failed to immediately deploy the lift-dump system after touchdown in accordance with company procedures, which negatively affected the airplane’s deceleration. 7. No evidence exists that reverted rubber or dynamic hydroplaning occurred. 8. If the captain had continued the landing and accepted the possibility of overrunning the runway instead of attempting to execute a go-around late in the landing roll, the accident most likely would have been prevented or the severity reduced because the airplane would have come to rest within the runway safety area. 9. Establishing a committed-to-stop point in the landing sequence beyond which a go-around should not be attempted for turbine-powered aircraft would eliminate ambiguity for pilots making decisions during time-critical events. 10. If, as a 14 Code of Federal Regulations Part 135 operator, East Coast Jets had been required to develop standard operating procedures and its pilots had been required to adhere to them, many of the deficiencies demonstrated by the pilots during the accident flight (for example, 87 NTSB Aircraft Accident Report inadequate checklist discipline and failure to conduct an approach briefing) might have been corrected by the resultant stricter cockpit discipline. 11.

[ANALYSIS] At 0938:27, the captain stated, “the sooner you get us there the better,” and then the first officer stated, “why don’t (they) just get us to the field?” These statements and those made earlier in the flight indicate that the pilots were impatient to land. Although no apparent reason existed for the pilots to feel rushed (for example, they landed 9 minutes ahead of schedule and no evidence was found that the passengers or the company were placing undue pressure on the pilots to land early on the day of the accident), they repeatedly expressed impatience with ATC and the weather radar displays. During postaccident interviews, other company pilots did not indicate that East Coast Jets pressured them to rush to get to the destination airport. The captain’s impatience during the flight was contrary to descriptions of his flying provided by other company pilots who described him as a serious and meticulous pilot. Specifically, the pilots described instances in which the captain carefully monitored weather during a flight and altered landing plans despite pressure from passengers.111 The captain could have reasonably opted to hold, divert, or attempt to land on runway 12 with a headwind. However, the captain’s focus on completing the flight degraded his attention to the changing weather situation and prevented him from recognizing that alternatives to landing on runway 30 were available. At 0938:50, the captain stated that the Approach checklist was complete, and, 1 second later, the first officer responded, “approaches are done,” even though he had been interrupted about 2 minutes 24 seconds before making this statement and had not completed the checklist. 111 One pilot described a flight with the accident captain in which a high-ranking official was a passenger. The flight was scheduled to land at Cedar Rapids, Iowa. Although the weather forecast was legal for IFR conditions, it was marginal. According to the pilot, the captain called FSSs repeatedly during the trip, determined that the weather was deteriorating, and decided to divert to the next scheduled stop at Des Moines, Iowa. The captain maintained his decision even though the travel arranger, who was also a passenger, came up to the cockpit, indicated that the high-ranking official was unhappy with the decision, and tried to talk the pilots into landing at Cedar Rapids despite the weather. NTSB Aircraft Accident Report Further, as the PM, the first officer should have verified that the checklist was complete, not the captain. At 0939:16, the CVR recorded the first officer trying to contact the FBO for nonessential reasons, such as asking about how to get fuel upon landing, with the captain’s approval at a time when he should have been completing the Approach checklist and monitoring the flight instruments. These calls were further interrupted by more critical communications with the captain, radio calls, and ATC and were not in accordance with 14 CFR 135.100, the “sterile cockpit rule,” which states that pilots may not make nonsafety-related radio calls during flight below 10,000 feet.

[ANALYSIS] The effective braking coefficients determined in the airplane performance study for the accident landing were substantially below those defined by the BCAR RWHS and AMJ 25X1591, assumed in the CAPS simulation, and underlying the wet runway landing distances provided in the BAe 125-800A AFM guidance material. The study also determined that the braking coefficients most representative of the actual performance of the airplane during the landing closely matched those calculated by a NASA friction expert using the CFME measurements made on the runway 2 days after the accident. The performance study also indicated that the total landing distances computed using the Section 25.109 braking coefficients 82 NTSB Aircraft Accident Report 83 can be significantly longer than those computed using the AMJ 25X1591 coefficients and provided in the AFM. For example, using the accident landing weight of 19,912 pounds, no wind, the OWA field elevation, the outside air temperature on the day of the accident, 140-psi tire pressure, and deceleration device deployment times, the airplane performance study indicated that the total landing distance using the AMJ 25X1591 braking coefficients was 3,338 feet and that the total landing distance using the Section 25.109 braking coefficients was 4,225 feet, which is 26 percent longer.137 (See table 4 in section 1.16.3.4.2) Assuming an 8-knot tailwind, the total landing distances were 3,792 and 4,928 feet, which is 32 percent longer, respectively. Adding the 15-percent safety margin recommended by SAFO 06012 to these distances, the required runway lengths with an 8-knot tailwind would be 4,361 feet using AMJ 25X1591 data and 5,667 feet using Section 25.109 data. Therefore, even if the accident flight crew had conducted an arrival landing distance assessment using the existing AMJ 25X1591-based AFM data (for either wind condition), it would have shown that the airplane could have stopped on the 5,500-foot runway with a safety margin of more than 15 percent. As shown, a landing distance assessment using Section 25.109 data would have indicated that the runway length was insufficient for landing with at least a 15-percent safety margin with an 8-knot tailwind. The airplane performance study indicated that the airplane would have exited the runway at between 23 and 37 knots and stopped between 100 and 300 feet beyond the runway end, but within the 1,000-foot runway safety area. The Section 25.109 calculations are consistent with current knowledge about wet runway braking performance, which is reflected in the engineering data used by the FAA to update regulations governing the calculation of accelerate-stop distances for wet, ungrooved runways138 and by the TALPA ARC in drafting new recommendations to require and support arrival landing distance assessments.

[ANALYSIS] A descent of 23,000 feet would require about 70 miles; therefore, to have been on a normal approach profile, the pilots should have started the descent about 20 miles further from OWA than they did. At 0927:48, the controller asked the captain to state his intentions and added that he could not provide a good recommendation at that time; however, the captain responded that it looked clear ahead for another 40 miles. At 0928:49, the captain stated to the first officer, “all I care is above ten [thousand feet] and we go fast so we can get around this…thing,” likely meaning that he wanted to maintain an indicated airspeed of more than 250 knots, which is the maximum airspeed allowed below 10,000 feet by 14 CFR 91.117, “Aircraft Speed.” At 0930:09, the captain stated during a conversation with the first officer, “good thing I didn’t tell ‘em it was gonna be a smooth ride huh? I looked at the radar and there wasn’t anything.” The first officer responded, “doesn’t it figure [weather] pops up right when we get here?” The captain continued, “what do you mean what are my intentions? Get me around this…storm so I can go to the field…I ain’t gonna turn around and go home.” About 0935, the pilots started the descent to 7,000 feet; however, according to the CVR recording neither pilot commanded the initiation of the Descent checklist. Even though the East Coast Jets Descent checklist is a silent checklist, it must still be initiated by command from the PF and be conducted and called complete by the PM. About 44 seconds later, the captain called for the Approach checklist, which, in accordance with company procedures, should have been preceded by an approach briefing that included the missed approach procedure, the runway conditions, and any “potential problems, such as weather.”110 However, according to the CVR, in response to the first officer’s calling for the approach briefing, the captain only replied, “it’s gonna be the ILS to three zero.” About 0937, the Rochester approach controller provided the first officer with weather information for OWA, which he stated was about 20 minutes old and indicated, in part, winds 320° at 8 knots (indicating headwinds for the intended runway). About 1 minute later, he added that light precipitation existed along almost the entire remaining route and that a couple of heavy cells were located within 5 miles of OWA. Despite this information indicating possible severe weather conditions, the CVR did not record either pilot verbally monitoring the wind information, which would have been displayed on each of the pilots’ electronic horizontal situation indicators and FMS displays. Data extraction of the captain’s FMS indicated that the instantaneous wind 12 seconds before landing was 195° at 17 knots, which would have resulted in a 5.6-knot tailwind.

[CONCLUSIONS > FINDINGS] Both pilots’ performance was likely impaired by fatigue that resulted from their significant acute sleep loss, early start time, and possible untreated sleep disorders, and fatigue might have especially degraded the captain’s performance and decision-making abilities when he had to decide while under time pressure whether to continue the landing or initiate a goaround. 19. Although the first officer took a prescription sleep aid for which he did not have a prescription the night before the accident, because of the short duration of its effects for most individuals, it is unlikely that the use of this medication degraded the first officer’s 88 NTSB Aircraft Accident Report performance at the time of the accident, which occurred about 12 hours after he took the medication. 20. Allowing civil aviation pilots who have occasional insomnia to use prescription sleep medications that have been proven safe and effective would improve these pilots’ sleep quality and operational abilities. 21. Educating and training pilots on fatigue-related issues could prevent pilots from operating flights while impaired by fatigue. 22. Formal guidance on how pilots can be treated for common sleep disorders while retaining their medical certification could help mitigate fatigue-related accidents and incidents. 23. The wet runway landing distances provided in aircraft flight manuals or performance supplemental materials that are based on the braking coefficients defined by the British Civil Air Regulations Reference Wet Hard Surface and Advisory Material Joint 25X1591 can be significantly shorter than the actual distances required to stop on some wet, ungrooved runways. 24. Title 14 Code of Federal Regulations Part 135 pilot-in-command line-check requirements are not adequate because they allow more than one required inspection to be conducted simultaneously and do not require that the line checks be conducted on flights that truly represent typical revenue operations; thus, the efficacy of line checks to promote and enhance safety is minimized, and pilots have limited opportunities to demonstrate their ability to manage weather information, checklist execution, sterile cockpit adherence, and other variables that might affect revenue flights. 25. Although the enhanced ground proximity warning system terrain database had not been updated to the most current version, the outdated database was not a factor in the accident. 26. A lightweight recording system conforming to European Organization for Civil Aviation Equipment ED-155, “Minimum Operational Performance Specification for Lightweight Flight Recorder Systems,” would have helped determine the flight crew’s actions during the landing and subsequent go-around attempt, including, but not limited to, whether they silently conducted checklists (partially or completely), which flap settings they selected, and how much braking effort they made upon landing. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the captain’s decision to attempt a go-around late in the landing roll with insufficient runway remaining.

[PROBABLE CAUSE] 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the captain’s decision to attempt a go-around late in the landing roll with insufficient runway remaining. Contributing to the accident were (1) the pilots’ poor crew coordination and lack of cockpit discipline; (2) fatigue, which likely impaired both pilots’ performance; and (3) the failure of the Federal Aviation Administration to require crew resource management training and standard operating procedures for Part 135 operators. 89 NTSB Aircraft Accident Report

[ANALYSIS] Both of these accidents might have been prevented if the pilots had committed to the landings or better understood where the committed-to-stop point was rather than attempting to go around with insufficient runway available to lift off and clear obstacles.115 The NTSB concludes that establishing a committed-to-stop point in the landing sequence beyond which a go-around should not be attempted for turbine-powered aircraft would eliminate ambiguity for pilots making decisions during time-critical events. Therefore, the NTSB recommends that the FAA require manufacturers of newly certificated and in-service turbine-powered aircraft to incorporate in their AFMs a committed-to-stop point in the landing sequence (for example, in the case of the Hawker Beechcraft 125-800A, once lift dump is deployed) beyond which a go-around should not be attempted. The NTSB further recommends that the FAA require 14 CFR Part 121, 135, and 91 subpart K operators and Part 142 training schools to incorporate the information from the revised manufacturers’ AFMs asked for in Safety Recommendation A-11-18 into their manuals and training. 2.3 Standard Operating Procedures SOPs are universally recognized as basic to safe aviation operations. Well-designed cockpit procedures are an effective countermeasure against operational errors, and disciplined compliance with SOPs, including strict checklist discipline, provides the basis for effective crew coordination and performance. SOPs should address, in part, checklist usage; radio 114 For example, see (a) Runway Overrun During Landing, Pinnacle Airlines Flight 4712, Bombardier/Canadair Regional Jet CL600-2B19, N8905F, Traverse City, Michigan, April 12, 2007, Aircraft Accident Report NTSB/AAR-08/02 (Washington, DC: National Transportation Safety Board, 2008) and (b) AAR-08/01. 115 The reports for these accidents, NTSB case numbers MIA06CA003 and ATL05CA131, respectively, are available online at <http://www.ntsb.gov/ntsb/query.asp>. Because airplane performance studies were not conducted for these accidents, it cannot be stated explicitly that the accidents would have been prevented if the pilots had committed to the landings. NTSB Aircraft Accident Report 64 communications; briefings; cockpit discipline, including sterile cockpit; stabilized approach criteria; CRM; and go-around/missed approach procedures. Operational data confirm the importance of strict compliance with SOPs for safe operations.116 For example, industry data show that pilots who intentionally deviated from SOPs were three times more likely to commit other types of errors, mismanage errors, and find themselves in undesired situations compared with pilots who did not intentionally deviate from procedures. According to AC 120-71A, in its study of controlled flight into terrain (CFIT) accidents, the Commercial Aviation Safety Team, which included FAA, industry, and union representatives, found that almost 50 percent of the studied CFIT accidents related to the flight crew’s failure to adhere to SOPs or the certificate holder’s failure to establish adequate SOPs.

[ANALYSIS] The Safety Board acknowledges that the runway marking and lighting were in accordance with FAA requirements, and does not consider them to be factors in this accident, except to the extent that they may not have provided the pilot with sufficient cues to cause him to be more attentive to the controller’s clearance. Runways 30R and 30L have complex approach lighting systems, which are especially visible at night. At the time of the accident, the white runway edge lights of runway 31 were operating at a dimmer setting than those of runways 30R and 3OL, which is standard practice at STL. The Safety Board believes that the dimmer lights on runway 31 were not sufficient to distract the pilot from his preconception that runway 30R was his intended departure runway. Finally, as the Cessna 441 pilot proceeded from taxiway Romeo into position on runway 30R, he entered the runway at an intersection 2,500 feet from the threshold. According to the AIM, an intersection clearance can be requested by the pilot or initiated by the controller. The Cessna pilot did not request an intersection takeoff, nor did the ground controller indicate that the pilot should expect an intersection departure, and the pilot should not have entered the runway at an intersection, without specific clearance to do so. While the Cessna 441 pilot’s entry onto the runway at an intersection should have been a final cue that his notion of being cleared to runway 30R was incorrect, the cue was apparently not sufficient to cause him to question his perception that he had been cleared to runway 30R. 2.2.3 Communications Effective radio communications between the Cessna 441 pilot and ATC were critical to establishing a mutual understanding of intentions. The ground controller’s multiple frequencies were congested with almost continuous communication, which resulted in several simultaneous transmissions in the 20 minutes before the accident. Additionally, there was some indication that the Cessna 441 pilot might have experienced communication radio difficulty. Specifically, the pilot complained about his communication radios during the inbound flight to STL, and several subsequent transmissions were garbled. Under these circumstances, it was especially critical for the pilot to ensure that effective communications were taking place. The Safety Board noted that the Cessna pilot did not sta!e the departure runway in any of his clearance readbacks. Although critical item readbacks have always been considered important in airborne operations, until recently, there was no requirement for critical item clearance readbacks for surface operz:ons. This omission was addressed in Safety 33 Recommendations A-95-33 and -34, which were issued, as mentioned in Section 1.18.5 of this report, during this investigation. In response to the recommendations, the FAA stated that it anticipates changing the AIM and Advisory Circulars 61-2 1 A, “Flight Training Handbook,” and 21-23B, “Pilots Handbook of Aeronautical Knowledge,” urging pilots to read back in full their runway assignment when operating at airports with more than one runway.

[ANALYSIS] During the on-scene investigation, several local pilots acknowledged that the proximity of runway 31 to the Midcoast ramp created a situation where pilots could inadvertently enter onto runway 3 1 without recognizing that they were on a runway. The Cessna 441 pilot had an airport diagram available in the cockpit, and may have referred to it during his outbound taxi. However, it is also possible that he believed the taxi route to runway 30R was obvious and thus did not pay much attention to the diagram. Even if he had referred to the airport diagram, he may have had difficulty discerning runway 31 on it, due to dim lighting in the cockpit, and the competing tasks that included taxiing the airplane and performing checklists. While runway 3 1 had markings, signage, and lighting consistent with FAA airport certification requirements, had several elements been slightly different, they might have triggered the pilot to question his notion that runway 30R was the assigned departure runway. These elements include runway width, displaced threshold, and runway markings and lighting. Runway 31 is 75 feet wide, which is typical of taxiways at STL. In contrast, runways 30R and 30L are 150 and 200 feet wide, respectively. At the time of the accident, a 1,838-foot displaced threshold was incorporated into the runway 3 1 marking/lighting scheme. The markings on the approximately 800-foot-long portion of runway 3 1 on which the Cessna pilot back-taxied consisted of a series of white arrows pointing toward the numbers. The runway 3 1 numbers were located at the end of the displaced threshold, near the intersection of runway 3 1 and taxiway November. The Cessna pilot’s taxi route did not go past the numbers. I 32 Along the displaced threshold, the runway lights had split red and white lenses, situated so that the white side of the lens was presented to the Cessna pilot as he back-taxied. This would have been a clue to the pilot that he was on a runway. However, the red side of the lens would have been visible to an airplane on approach for the runway, or to a pilot holding in position on runway 31 for departure. Because of the displaced threshold marking scheme, the Cessna pilot could not have seen the numbers for runway 3 1. Had he seen the numbers, the pilot might have been cued to question the controller as to the controller’s intentions. The Safety Board acknowledges that the runway marking and lighting were in accordance with FAA requirements, and does not consider them to be factors in this accident, except to the extent that they may not have provided the pilot with sufficient cues to cause him to be more attentive to the controller’s clearance. Runways 30R and 30L have complex approach lighting systems, which are especially visible at night.

[CONCLUSIONS > FINDINGS] There was no mention of the occasional use of runway 31. 10. The controller clearly referenced runway 3 1 in two separate transmissions. In both cases, the pilot acknowledged the clearance, but did not read back the runway assignment. Had the controller used more precise phraseology in the issuance of the initial taxi clearance, the Cessna 441 pilot may have noted the proper departure runway. 11. Had the Cessna 441 pilot volunteered, or had the controller requested, 48 confirmation of the assigned runway, the pilot’s error may have been detected and the accident prevented. 12. Air traffic control personnel were not able to maintain visual contact with the Cessna 441 after it taxied from the well-lighted ramp area into the runway/taxiway environment of the northeast portion of the STL airport. 13. An operational ASDE-3, particularly ASDE-3 enhanced with AMASS, could be used to supplement visual scan of the northeast portion of the STL airport surface. 14. The MD-82 flightcrew stated that they did not observe any external lights on the Cessna 441 before impact. When the Cessna 441 was in position for departure on runway 30R, the most conspicuous exterior lighting was directed forward, and, with the possible exception of wing anticollision/strobe lights, would not have been visible to the MD-82 flightcrew. 15. It is likely that the wing anticollision/strobe lights were not operating when the collision occurred. 16. Pilot training for surface movement can be improved in both air carrier and general aviation areas. 3.2 Pm bable Cause The National Transportation Safety Board determines that the probable cause of this accident was: the Cessna 441 pilot’s mistaken belief that his assigned departure runway was runway 30R, which resulted in his undetected entrance onto runway 30R, which was being used by the MD-82 for its departure. Contributing to the accident was the lack of ATIS and other ATC information regarding the occasional use of runway 31 for departure. The utilization of an operational ASDE-3, and particularly ASDE-3 enhanced with AMASS, could have prevented this accident. 49

[ANALYSIS] Had the equipment been operational, it would have displayed the Cessna 441’s position at the intersection of taxiway Romeo and runway 30R for about 3 minutes before the collision. 3 30 The circumstances of this accident indicate that the pilot of the Cessna 441 unintentionally deviated from the taxi clearance he received from the ground controller and taxied onto the active runway being used by the MD-82. The Safety Board’s investigation examined possible reasons why the Cessna pilot might have believed that runway 30R was his departure runway. 2.2 Cessna 441 Pilot Performance The Safety Board believes that several personal factors may have contributed to the Cessna 441 pilot’s deviation from ATC instructions. According to the pilot’s wife, the accident occurred at a time of night when the pilot normally went to sleep, and he may have been tired. Company personnel reported that such late trips were unusual. Although the pilot’s workrest cycle is not consistent with chronic sleep loss, the fact that he was operating during a period in which he was normally at rest may have had some effect on his performance and level of attentiveness. Additionally, the pilot of the Cessna 441 had commented that it was going to snow in Iron Mountain that night. Midcoast personnel stated that the pilot seemed anxious to go home, a behavior that they considered normal among pilots at that time of night. The combination of the time of day and his desire to return home before the weather deteriorated may have contributed to the mistaken actions of the Cessna 441 pilot, who was generally described in positive terms for his cautious and safe attitude. 2.2.1 Scenario: Pilot Became Lost During Airport Ground Operations The Safety Board considered the possibility that the pilot intended to take off from runway 3 1, as directed, but became lost on the airport and ended up in position to take off on the wrong runway. However, the pilot did not indicate confusion in his radio responses to the taxi clearance, and radar data indicated no hesitation in his taxi route. (See Appendix E.) The passenger who chartered the Cessna 441 to STL before the accident flight reported that on a previous flight when the pilot of the Cessna 441 had become uncertain of his position at an airport, he had stopped taxiing until he determined his location. She also indicated that it was the pilot’s habit to taxi with the airport diagram in front of him. The current STL airport diagram approach chart was located in the cockpit area of the Cessna 441 wreckage. Additionally, the pilot’s flight logbook revealed that he had landed once before at STL.

[ANALYSIS] She also indicated that it was the pilot’s habit to taxi with the airport diagram in front of him. The current STL airport diagram approach chart was located in the cockpit area of the Cessna 441 wreckage. Additionally, the pilot’s flight logbook revealed that he had landed once before at STL. The previous flight was a daytime operation and occurred approximately 10 months before the accident. 2.2.2 Alternate Scenario: Pilot’s Pleconceplion that Runway 30R was his Ihparture Runway The evidence indicates that it was unlikely that the pilot was lost, but rather that he had a preconception that he would be departing on runway 30R and thus did not register the ground controlle.-‘s clearance to runway 3 1. Several situational cues may have reinforcec ‘ie Cessna 441 pilot s preconception that runway 30R was his assigned departure runway. ? he Cessna 441 pilot had landed on runway 30R about 18 minutes before he received the taxi 31 clearance to runway 3 1 for his departure. The “quick turnaround” nature of the flight may have added to the Cessna pilot’s belief that he would be departing on runway 30R. Also, from the time he approached STL for landing until he taxied out for takeoff, all traffic had landed and departed on runways 30R and 30L. The ATIS that was current during the time the pilot operated in the STL area listed runways 30R and 30L as the active runways for arrivals and departures at STL. The STL controllers did not typically list runway 31 as an active runway on the ATIS, as runway 31 was only occasionally used as a departure-only runway. Also, the STL controllers did not typically treat runway 3 1 as if it were an active runway; for example, when the Cessna 44 1 pilot cleared runway 30R on his inbound flight, his taxi clearance to the Midcoast ramp did not include a clearance to cross runway 31. The Safety Board believes that if runway 31 had been referenced as a runway for occasional general aviation departures on the ATIS broadcast, the pilot may have been more attentive to the controller’s taxi clearance and runway assignment. Another situational cue that could have reinforced the Cessna 441 pilot’s notion that runway 30R was his departure runway was the fact that when he began to taxi outbound from the Midcoast ramp on taxiway Whiskey (150 feet long), he almost immediately encountered runway 3 1, unlike the more typical airport layout in which a ramp exit leads to a parallel taxiway en route to the runway. During the on-scene investigation, several local pilots acknowledged that the proximity of runway 31 to the Midcoast ramp created a situation where pilots could inadvertently enter onto runway 3 1 without recognizing that they were on a runway.

[FINDINGS] There was a thin layer of slush on the runway at the time of the landing. The pilot was aware of the weather and runway conditions prior to the landing approgch. The aircraft operating manual contained data which showed that the aircraft could not have stopped safely on runway 23. The aircraft touched down at a higher-than-normal groundspeed which further degraded the runway braking coefficient of friction. The pilot failed to iminediately recognize the insuffictent b-aking action and delayed his atterapted go-around until the alreraft was about 1,200 feet from the end of the runway. There was insufficient runway remaining for the go-around. There was no physical evidence to Indicate that the aircraft was in a hycroplaning condition when it overran the runway. The crash was impact-survivable for the passengers since the decalerative forces were within the limits of human tolerance. The flightcrew died as a result of blunt impact trauma. ~20- There was an intense posterash fire. The passengers died as a result of thermal injuries sustained in the posterash fire, The unsuccessful attempted go-around increased the severity of impact forces and the possibility of a posterasn fire and fatal injuries. The accident may have been survivable had the pilot continued to decelerate the aircraft. 20. The pilot exercised poor judgment in attempting the landing under the conditions that existed. 21. The AFM did not adequately emphasize the required landing distances under wet and icy runway conditions, 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of the accident was the pilot's attempt to land on a slush covered runway with insufficient Stopping distance available, and his delayed initiation of a go-around which resulted in there being insufficient runway available to complete the maneuver successfully. Contributing to the accident was the lack of adequate emphasis in the manufacturer's aireraft flight manual regarding the required aircraft landing/stop distances under wet and icy runway conditions. 4, RECOMMENDATIONS As a result of its investigation and previous overshoot accidents with the Citation, the Safety Board made the foliowing recommendations to the Federal Aviation Administration: Require Cessna to include in the appropriate sections of all Citation aircraft flight menuals the portion on page IV-3 of the manufacturer's aircraft operating manual which pertains to landing on slippery runways. (Class Il, Priority Actior.) (A-81-65) Require «‘essna to include in the appropriate sections of all Citation aircraft flight manuals a warning that solid ice, snow, or slush corrected landing distances may not be adequate in operations. (Class II, Priority Action) (A-81-6$) Through advisory circulars and/or operations bulletins, emphasize and reinforce in the training curricula for at least all turbojet initial and recurrent phases the liinitations and the hazards that may be encountered when landing on slippery runways. (Class 1], Priority Action) A-81-67) Revie'y and require revisions as appropriate of :nanufacturer's aircraft flight manuals to include sufficient slippery runway condition correction factor information or require an appropriate warning that landing distances under slippery runway conditions are unknown. (Class Il, Priority Action) (A-81-68) BY THE NATIONAL TRANSPORTATION SAFETY FOARD /s/ ELWOOD T.

[ANALYSIS] The tires exhibited normal wear which indicates that viscous hydroplaning wes not a factor. In any event, the runway condition would have reduced the effectiveness of the braking action considerably, and the higher-than-normal groundspeed at touchdown made the increased runway distance factors in the aircraft operating manual completely inadequate. Furthermore, the 5,250 feet required because of the icy runway condition exceeded the total length of runway 23, and the pilot should have known this information. . In view of the pilot's ensiderable experience in the Citation, he should have been thoroughly familiar with its performance and characteristics, and therefore he exercised poor judgment in attempting the landing. It fs possible that the pilot intended to land for the purpose of checking the braking action with the subsequeat intention to go around, a procedure alluded to by the chief pilot as having been previously used. The decision to attempt .a go-around apparently was delayed because he failed to immediately recognize the insufficient braking action and failed to consider the drag prc-duced by the slush on the runway. Since « landing was made, the pilot should have been aware of the fact that a timely decision to execute a go-around would be absolutely necessary under the circumstances and that any delay would jeopardize the success of such a maneuver. Since a delay occurred, a decision by the pilot not to attempt a go-around bit to continue braking probably would have decreased the severity of the impact and may have resulted in « survivable accident. The Safety Board notes that company policy did not preclude the pilot from landing under the existing conditions nor was any guidance given in the company operations manual. The Safety Board recognizes that this information was provided to the pilot in his training program and was also available in the operating manual whici. was maintained by the company. We believe that this information in addition to his training and experience should have been sufficient to make a proper decision. However, in the Safety Board's opinion, it is essential that this information be consolidated and be presented clearly In the AFM and emphasized In training so that a pilot thoroughly understands :-. > information and its implications. 3. CONCLUSIONS Findings The flighterew was properly certificated and qualified for the flight. They were familiar with the Mercer County Airport. There was no medical evidence indicating any impairment in the flightcrew's physical ability to operate the aircraft, The aircraft was certificated and maintained in accordance with approved maintenance procedures. The right engine anti-ic> fail light discrepancy did not cause or contribute to the accident; the source of the discrepancy could not be determined. There was no other evidence of a mechanical failure or malfunction of the aircraft which either would have caused or contributed to the accident, The weather forecasts covering the period of the flight were Substantially correct, There was light *o moderate rime icing in clouds below 20,000 feet; there was no significant wind shear or turbulence present below 20,000 feet in the vicinity of the airport. There was a thin layer of slush on the runway at the time of the landing.

[ANALYSIS] Although the glide slope touchdown point is 1,057 feet beyond the threshold of the runway, the tailwind component may have caused the alreeaft to touch down farther on the runway than the pilot Intended. After the touchdown, thrust would have been at idle, brakes would have been applied, and the spoilers would have been deployed. A perceived lack of deceleration would have prompted the pilot to execute a go-around a3 soon as he realized that he could not stop the aircraft on the cunway. In this case, he failed to promptly recognize the insufficient braking action and therefore, he did not initiate a timely go-around. A go-around becomes marginal at low speeds because of the length of runway required to accelerate to liftoff speed. The witnesses were consistent in their reports that engine thrust was increased between 1,000 and 1,200 feet from the departure end of the runway. They stated that the eircraft rotated to a takeoff attitude before crossing the end cf the runway, but that it did not get airborne. According to the manufacturer, the minimura speed at which the nosewheel can be raised off the runway is about 81 KIAS, and the aircraft can become airborne at a minimum speed of 91 KIAS. Therefore, based on the manufacturer's information and the witnesses! reports, the aircraft could have departed the runway at a speed between 81 and 91 knots. However, since the aircraft did not become airborne, it is evident that it did not reech sufficient speed for liftoff. The pilot might have achieved the minimum liftoff speed if he had applied full thrust between 1,400 and 1,800 feet from the end of the runway; he would have had between 9 to 15 seconds after he touched down to make the decision. The drag produced by the slush would have extended the distance needed to accelerete and, thus shortened the available decision interval. Based on the icy runway condition correction factor of 100 percent in the alreraft operating manual, the aircraft would have required about 5,250 feet to stop. This calculated distance inchides the increased landing distance with a 10-knot tailwind. However, the aircraft's computed groundspeed at touchdown was 111 knots with the 9-knot tailwind component, resulting in a speed 21 knots above the minimum dynamic bydroplaning initiation groundspeed of 90 knots. Since the icy runway condition correction factor of 100 percent is based on 90 knots, it Is not adequate; when touchdown groundspeeds are greater. ‘Therefore, under the foregoing conditions, the aircraft's required lar.ding distance would have been greater than 5,250 feet. The weather conditions that existed ui Bluefield were known to the pilot since he had briefed himself on the weather at the NWS office before departure.

[ANALYSIS] Since the icy runway condition correction factor of 100 percent is based on 90 knots, it Is not adequate; when touchdown groundspeeds are greater. ‘Therefore, under the foregoing conditions, the aircraft's required lar.ding distance would have been greater than 5,250 feet. The weather conditions that existed ui Bluefield were known to the pilot since he had briefed himself on the weather at the NWS office before departure. He had adequate Information available to him from which to decide whether or not a safe landing could be made at the alrport. If he had inquir.d about the condition of the runway at Bluefield before departure, he would have been aware of the adverse condition. Moreover, he recelved the runway condition reports from Roanoke Approach Control and the FSS. He had considerable experience In the alreraft and was familiar with the airport even though 4 months had elapsed since his last flight into Bluefield. The fact that he was ~18- aware of the critical length of the runway was demonstrated by his execution of the circling approach to runway 5 following the first ILS approach to runway 23, It is possible, however, that he may have been misled by the report that snowplows were on the runway when he was approaching the airport, and he may have aniticipatec an improved runway condition following the plowing operation. Also, the freshly pluwed runway which he observed during the circling approach to runway 05 may have appeared to be in better condition than it was as was later reported by another pilot following a subsequent plowing. These factors mey have influenced the pilot's decision to execute the second approach to runway 23. Although a wind of 070° at 10 knots would have reduced the landing distance, 4 safe landing probably could not have been made on runway 05. If the pilot had completed his landing approach to runway 05 to take advantage of a §8-knot headwind, the aircraft's groundspeed at touchdown would have been about 93 knots. In order for the e!reraft to have had a 90-knot groundspeed at touchdown, its landing weight would have had to be reduced by 400 pounds, to a landing weight of 7,900 pounds if it were made in a no-wind condition. Since the empty weight of the aircraft was 6,709 pounds and the tote! occupant weight was $19 pounds, only 281 pounds of fuel could have been on board for a landing. The total amount of fuel on beard at the time of the landing was 2,431 pounds. The Safety Board could not positively determine whether the aircraft encountered hydroplaning because there was no evidence on the runway following the second plowing to indicate that the wheels failed to spin up. The tires exhibited normal wear which indicates that viscous hydroplaning wes not a factor. In any event, the runway condition would have reduced the effectiveness of the braking action considerably, and the higher-than-normal groundspeed at touchdown made the increased runway distance factors in the aircraft operating manual completely inadequate.

[ANALYSIS] The terrain below the airplane dropped some more and then rose slightly as the airplane continued its descent. 52 The FPA announced the 700- and 600-foot callouts before the first officer’s response. NTSB/AAB-02/03 35 decreasing.53 At this point, the airplane was at an altitude of 8,100 feet and was 1.2 miles from the runway. Title 14 CFR 91.175(c)(3) states that at least one specified visual reference for the intended runway needs to be “distinctly visible and identifiable to the pilot” for the airplane to operate below the MDA. The captain’s statement “where’s it at?” indicated that he did not see any of these visual references. Even if the first officer had the runway in sight at this point, the captain, as the flying pilot, should not have been relying on the first officer for directional guidance during the visual transition from the instrument approach to the landing. Radar data indicated that, about 1901:47, the airplane stopped turning to the right and began turning to the left. The Safety Board regards this maneuver as the first clear indication that the captain may have seen the runway after the MDA. About 1901:49, the airplane was at an altitude of 8,000 feet, was about 0.9 miles north of the runway, and was descending at a rate of 900 feet per minute. At that point, the airplane was banked 10º left wing down. The airplane’s bank angle continued to increase.54 The local controller saw the airplane emerge from a snow shower (at a low altitude and west of the runway extended centerline) and rapidly enter a steep left bank. On the basis of the ATC voice recordings and information from a postaccident interview, the Board estimates that the airplane crashed about 5 seconds after the controller saw the airplane. At the time of the accident (1901:57), the airplane was banked more than 40º left wing down, and the left wing tip was the first airplane part that struck the terrain. The CVR contained no indication regarding why the airplane was turning so steeply to the left. It is possible that the captain saw the runway or the highway and was making a steep turn to align the airplane with the runway without a substantial overshoot.55 It is also possible that the captain was starting to see terrain that had been obscured by the darkness and the weather conditions and was banking the airplane to avoid the terrain. The left and right main landing gear were found in their wheel wells. However, no evidence on the CVR and from the CVR Sound Spectrum Study indicated that the gear was being raised for a missed approach. In addition, the airplane manufacturer stated that a warning horn would sound if the gear were being raised while the flaps were in the landing configuration.

[ANALYSIS] However, these actions were contrary to spoiler information in the Gulfstream GIII Flight Manual, which indicates that the spoilers are not to be extended with flaps in the landing configuration (39º) or with the landing gear extended and that the high-pressure rpm power setting on final approach should not be below 64 percent N2 to meet FAA-required go-around standards. About 1901:34, the airplane’s attitude started to increase in the right-wing-down direction. (The airplane should have been turning to the left to align with the runway.) The airplane passed the missed approach point about 1901:36 at an altitude of about 8,300 feet, 485 feet above field elevation rather than the specified 2,385 feet above field elevation (see figure 3). The first officer was required to call out, “missed approach point, runway in sight” or “missed approach point, runway not in sight,” and the captain was required to announce his intentions. However, the CVR did not record either of these callouts. About the same time as the airplane passed the missed approach point, the captain asked, “where’s it at?” This statement suggests that the captain had not identified, or had lost visual contact with, the runway. At this point, the captain should have abandoned the approach, especially because the airplane was close to the ground in mountainous terrain.51 The first officer stated, “to the right,” about 6 seconds after the captain’s query.52 Again, it is not apparent what the first officer could see from the cockpit when he made this statement. After the first officer’s response, the captain stated, “to the right,” but calculations from radar data indicated that the airplane’s right bank angle was 50 The CVR indicated that the captain had called for the landing gear and landing flaps about 1859:30 and 1859:34, respectively. (The airplane was at an altitude of 12,200 feet when the captain called for the landing gear.) The first officer’s statement, “three greens,” about 1859:46 indicated that the landing gear was in the down and locked position. 51 The captain and the first officer should have realized the proximity of the airplane to the ground because the flight profile advisory (FPA) unit had announced the 1,000-, 900-, and 800-foot callouts before the captain’s question about the location of the runway. Although the airplane was less than 500 feet above airport elevation at the time of the captain’s question, the ground proximity warning system (GPWS) and FPA 500-foot altitude were not announced then because, according to the Airplane Performance Study, the airplane was flying over a river valley at that point and the terrain elevation directly below the aircraft was 7,600 feet—over 700 feet lower than the airport elevation. The terrain below the airplane dropped some more and then rose slightly as the airplane continued its descent. 52 The FPA announced the 700- and 600-foot callouts before the first officer’s response.

[ANALYSIS] Also, the airplane was now operating below the 10,200-foot MDA without any indication, according to the CVR, that either pilot had made visual contact with the runway or its environment. The CVR indicated that the first officer did not verbally challenge the captain, and radar data showed that the captain did not correct the descent or initiate a missed approach. About 1900:43, the captain asked the first officer whether he could see the runway, but the CVR recorded an unintelligible statement made by the first officer about 2 seconds later. About 1900:46, the captain asked the first officer whether he could see the highway. (The highway, as viewed from the approach, was located slightly to the right of the runway extended centerline and thus would have been more easily visible on the first officer’s side of the cockpit.) The CVR recorded the first officer’s statement “see highway” 1 second later, but this statement does not clearly indicate whether he actually had the highway in sight. Because the first officer provided no specific NTSB/AAB-02/03 33 information about the highway’s location, it is possible that he was repeating the captain’s words while looking for the highway. The captain made no statements about this time to indicate that he had established visual reference with either the runway or the highway. The airplane began to descend again about 1900:49 at a rate of 2,200 feet per minute. The local controller noticed that the airplane had prematurely descended below the 10,400-foot step-down altitude and asked the flight crewmembers whether they had the runway in sight. About 1900:51, the first officer and the captain stated, almost simultaneously, “affirmative” and “yes now yeah we do,” respectively. These statements were communicated only to the other pilot, but, about 1 second later, the first officer informed the controller that the runway was in sight. Radar data indicated that the airplane was at an altitude of 9,750 feet at the time but that it had not started maneuvering toward the airport. Evidence indicated that, at that point, the flight crew probably did not have the runway in sight or had it in sight only briefly. Specifically, the CVR did not record any previous independent indication from either flight crewmember that he had visually identified the runway; both pilots stated that they saw the runway only after being queried by the controller. Also, the CVR did not record any further discussion throughout the rest of the flight that would be consistent with a flight crew that could see a runway. Neither flight crewmember commented about the threshold or its markings or lights; the runway end identifier lights; the precision approach path indicator; the touchdown zone or its markings or lights; or the runway or its markings or lights—one of which needed to be in sight to make a landing in accordance with 14 CFR 91.175(c)(3). In addition, the local controller stated that she did not see the airplane when the pilot reported the runway in sight.

[ANALYSIS] The left and right main landing gear were found in their wheel wells. However, no evidence on the CVR and from the CVR Sound Spectrum Study indicated that the gear was being raised for a missed approach. In addition, the airplane manufacturer stated that a warning horn would sound if the gear were being raised while the flaps were in the landing configuration. The CVR did not record the sound of this warning horn. 53 While the airplane was still banked to the right, the GPWS and FPA announced the 500-foot altitude callout. 54 As the airplane’s left bank angle was increasing, the FPA announced the 400-, 300- and 200foot altitude callouts, and the GPWS announced the 200-foot callout, two sink rate alerts, and one bank angle alert. 55 Radar data showed that the airplane was substantially to the right of the runway extended centerline. NTSB/AAB-02/03 36 Summary of the Flight Crew’s Performance During the final 2 minutes of the approach, the flight crewmembers were apparently focusing more of their attention outside, rather than inside, the cockpit as they tried to locate the runway and the highway. As a result, the captain continued flight below the authorized MDA after failing to establish or maintain visual contact with the runway. The first officer did not challenge the captain’s actions. In addition to their numerous errors during the instrument approach procedure, the flight crewmembers demonstrated poor crew coordination during the accident flight. Specifically, the captain and the first officer did not make required instrument approach callouts,56 the captain provided an incomplete approach briefing, and he did not follow Gulfstream’s procedures regarding in-flight spoiler operation and minimum engine power during an approach. The flight crewmembers should have abandoned the approach because the airplane descended below the MDA without an adequate visual reference of the runway. Also, the flight crew should have considered diverting to an alternate airport after receiving information about the deteriorating visibility along the approach course and the three reports of missed approaches. Operational and human factors that may have played a role in the flight crew’s decision to continue the approach to an intended landing are discussed in the next section. Visual factors affecting the flight crew’s ability to see and avoid the terrain are discussed later. Operational and Human Factors The flight crew was dealing with various sources of operational pressure during the accident flight. These sources were the airplane’s late departure from LAX, the ASE nighttime landing curfew, and the charter customer’s strong desire to land at ASE. Because of the late arrival of the passengers (including the charter customer), the airplane was not able to depart LAX until 1711—41 minutes later than scheduled. The late departure delayed the flight’s estimated arrival at ASE to 1846—only 12 minutes before the airport’s landing curfew.

[ANALYSIS] Neither flight crewmember commented about the threshold or its markings or lights; the runway end identifier lights; the precision approach path indicator; the touchdown zone or its markings or lights; or the runway or its markings or lights—one of which needed to be in sight to make a landing in accordance with 14 CFR 91.175(c)(3). In addition, the local controller stated that she did not see the airplane when the pilot reported the runway in sight. Further, the airplane would have had to make a left turn to align with the runway, but radar data showed that the airplane was turning slightly to the right (2.5º right wing down). The airplane passed the step-down fix located 9.5 DME south of the Red Table VOR about 1901:00 at an altitude of 9,500 feet, 900 feet below the specified minimum altitude (see figure 3). Because the airplane was still in instrument meteorological conditions, the first officer should have announced this altitude deviation to the captain, but the CVR did not record any such callout. The 1853 Automated Surface Observing System hourly observation indicated that the lowest cloud base was near 9,315 feet. Once the airplane had descended below this altitude, the pilots were likely encountering visibilities of about 2 miles in light snow showers. The airplane was 2.7 miles from the airport at this time. About 1901:13, the CVR recorded the first officer’s statement, “…to the right is good.” Radar data indicated that the airplane was at an altitude of 9,000 feet. It is not apparent what the first officer could see from the cockpit when he made his statement because the runway would have been to the left of the nose of the airplane. It is possible that the first officer could have seen the runway at this point; however, the captain did not NTSB/AAB-02/03 34 verbally acknowledge the first officer’s directional guidance, and radar data indicated that the airplane did not make a turn to the right. The airplane continued on its heading, which would have still positioned the airplane to the right of the runway. About 1901:21, the CVR recorded a sound, which continued for 9 seconds, that was consistent with the airplane’s configuration alarm. This warning indicated that the captain had deployed the spoilers after the landing gear had been extended and the final landing flaps had been selected.50 Also, the CVR Sound Spectrum Study determined that, while the spoilers were deployed, engine power was set to 55 percent N2. The captain likely extended the spoilers and reduced engine power to increase the airplane’s rate of descent to get below the snow showers and visually acquire the runway. However, these actions were contrary to spoiler information in the Gulfstream GIII Flight Manual, which indicates that the spoilers are not to be extended with flaps in the landing configuration (39º) or with the landing gear extended and that the high-pressure rpm power setting on final approach should not be below 64 percent N2 to meet FAA-required go-around standards.

[ANALYSIS] However, the presence of this passenger in the cockpit, especially if it were the charter customer, most likely further heightened the pressure on the flight crew to land at ASE. The operational pressure on the flight crew probably resulted in the crew’s intent to continue with its original plan to land at ASE. FAA Advisory Circular 60-22, “Aeronautical Decision Making,” indicates that pilots, particularly those with considerable experience, try to complete flights as planned, please passengers, and meet schedules, which can compromise safety and impose an unrealistic assessment of piloting skills under stressful conditions. Also, human performance researchers have noted that pilots tend to adhere to their original plan of action, which interferes with critical analysis processes that are needed to adequately reevaluate the suitability of the original plan and explore an alternate course of action. As a result, “plan continuation errors” occur; that is, pilots elect to continue with an original plan of action despite the presence of cues suggesting that the course of action needs to be modified.58 In addition, research has demonstrated that individuals, when faced with a choice between alternatives, generally seek out information that confirms a chosen hypothesis and ignore or fail to fully 57 As previously indicated, 14 CFR 135.100 did not prohibit the captain from allowing this passenger to sit on the jumpseat and from conversing with him because the airplane had not yet reached the sterile cockpit altitude of 10,000 feet msl. 58 J. Orasanu, “Plan Continuation Errors: A Factor in Aviation Accidents?” Proceedings of the Fourteenth Triennial Congress of the International Ergonomics Association and Forty-forth Annual Meeting of the Human Factors and Ergonomics Society (Santa Monica, CA: Human Factors and Ergonomics Society, 2000). NTSB/AAB-02/03 38 consider contradictory information, particularly when workload is high and time constraints are imposed.59 In this accident, the flight crew focused on cues that supported the goal of landing at ASE, such as occasional breaks in the clouds and the report of one airplane (N900MF) that had landed at the airport without incident. By focusing on such cues, the crew did not adequately consider information that supported a change of plans, such as the deteriorating visibility, the three reports of missed approaches (the last of which, involving N527JA, occurred after the accident airplane passed the Red Table VOR), and the failure to establish and maintain visual contact with the runway environment as the approach proceeded. Instead, the flight crewmembers might have viewed this information as a barrier to achieving their intended goal of landing at ASE, causing them not to recognize the increasing evidence supporting the need for an alternate course of action. Visual Factors When the VOR/DME-C instrument approach procedure was first established at ASE in December 1988, the procedure was not authorized at night.

[ANALYSIS] This information may be omitted if it is contained in the ATIS broadcast and the pilot states the appropriate ATIS code.” NTSB/AAB-02/03 32 visibility to the north should have alerted the flight crewmembers that they might also need to execute a missed approach because of the weather. However, the CVR did not record any discussion between the crewmembers at this point regarding a possible missed approach. The final approach segment required the flight crew to maintain 12,200 feet until passing ALLIX, the final approach fix located 6 DME south of the Red Table VOR; 10,400 feet until passing a step-down fix located 9.5 DME south of the Red Table VOR; and 10,200 feet until passing the missed approach point located 11 DME south of the Red Table VOR. (See figure 1.) The Airplane Performance Study indicated that the airplane crossed ALLIX about 1859:40 at an altitude of 12,100 feet, 100 feet below the minimum specified altitude (see figure 3), and at a speed of about 125 knots. After passing ALLIX, the airplane maintained a descent rate of about 2,200 feet per minute and an airspeed of about 125 knots. When the airplane reached altitudes of 11,200, 10,700, and 10,300 feet, Avjet’s Operations Manual required the first officer to call out, “1000 to go [until landing minimums],” “500 to go,” and “approaching minimums,” respectively. However, the CVR did not record any of these callouts. About 1900:22, when the airplane was at an altitude of 10,400 feet and was about 4.4 miles north of the airport, the captain stated, “okay…I’m breaking out,” which was the first indication that the captain might have made visual contact with the ground. About 5 seconds later, the captain asked the controller whether the lights were all the way up. Postaccident interviews and CVR evidence determined that, by this time, the ground controller had set the airport lights at their highest intensity; thus, the captain’s query could suggest that he did not have the runway or its environment in sight. It is also possible that the captain could have perceived the lights as dim because of an obscuration and wanted to make sure the lights remained visible. The airplane leveled off about 1900:39 at an altitude of 10,100 feet and about 3.7 miles north of the airport. The CVR did not record any announcement by the first officer that the airplane was 300 feet below the minimum step-down altitude of 10,400 feet. Also, the airplane was now operating below the 10,200-foot MDA without any indication, according to the CVR, that either pilot had made visual contact with the runway or its environment. The CVR indicated that the first officer did not verbally challenge the captain, and radar data showed that the captain did not correct the descent or initiate a missed approach.

[ANALYSIS > THE ACCIDENT] The captain stated that he still believed that the aircraft would Jand on the runway, although closer to the threshold than he had planned. Airspeed was maintained at or near reference speed until the point where power was reduced about 50 feet above the runway. At that point, the descent rate increased rapidly when the thrust was reduced to idle. Even thvugh the control yoke was probably pulled aft in an alieimpt to maintain the approach path, without power the airspeed decrcased rapidly and the deseent rate increased rapidly because the aircraft had insufficient thrust in relation to drag to reach the runway. Therefore, the aircraft landed short of the runway because the captain prematurely reduced the thrust. There are several reasons why the captain arrived at a point in this approach where he mistakenly reduced thrust and landed short. Of these reasons, the one of major concern to the Safety Board was the manner in which the approach was flown. The Safety Board believes that the captain's failure to fly a standard, approved pattern directly contributed to the final outeome. It was apparent from the captain's statements that he was ecneerned about the short runway, and that he intended to touch down before the company-prescribed touchdown point of 1,000 feet. The captain's training in both the DC-10 and B-727 aircraft and flight manual procedures emphasized the need to plan a pattern for a touchdown aim point of 1,000 feet beyond the threshold of the runway. Admittedly, the length of the runway at Yap (4,820 feet) is comparatively short; however, Bt pi ee Spare ite oe PNY tS pte RS Bale ABI Fm be tee ee the stopping procedures anc certification data for the aircraft insure a safe landing if recommended pattern procedures are followed. The Safety Board believes: that the captain was ignoring these criteria and was concerned about the short length of the runway; therefore, he planned to land about 300 feet rather than 1,000 feet beyond the runway threshold. The approach to Yap was not typical of the type previously flown by the capiain. The fly-by procedure to cheek the runway placed the aircraft in an abnormal position on the downwind leg of the pattern. Once the fly~by was completed, however, the captain was required to establish a normal bese leg and final approach. In this case, the captain did not regain the proper altitude for a normal base leg; instead he turned for the final approach about 1.5 miles from the runway at only 250 feet above the runway slevation instead of being stabilized on the final at 500 feet as recommended in the approved flight manual. If he had turned on the final approach at the same distance but at the proper altitude of 500 feet, he would have been on a normal 3° approach slope angle to the 1,000-foot aim point. However, the low base~-leg aititude and turn to the final approach required a flat approach slope angle of about 1.5° and a low rate of descent.

[ANALYSIS > THE ACCIDENT] Similarly, it appears narrower for the remaining length because of the grass growing through the runway surface. The classic problem of runway width causing illusions pertains to the fact that the pilot us-s the apparent convergence angle of the runway edges in perspective to estimate length. Increasing or decreasing the distance between the lines can create illusions of shortening or lengthening of the pilot's perception of the runway length. The wider the runway is from that normally encountered by the pilot, the shorter it appears; but a wider runway also can cause the pilot to think he is lower than his actual height above the runway. In the case of the Yap runway, the width is ill-defined; however, it tends to give the i?lusion of being Jonger than its actual length, because the narrower width toward the far end of the runway increases the apparent convergence. Regardless of the possibilities of illusions because of the Yap runway condition, the Safety Board cannot conclude that this factor contributed to the low, flat approach flown by the captain or to his premature reduction of thrust. The Safety Board believes the pilot was not affected by any of these illusions because he stated that he aimed for about a 300-foot touchdown point rather than the presvribed 1,000-foot point. His aim was actually quite accurate because, if he had not reduced power when he did, he probably would have touched down at or very near his aim point. Therefore, the Safety Board does not believe that visual illusions were a factor in this accident. The captain engaged in and permitted distracting conversations in the cockpit during the downwind portion of the approach. The taking of pictures and discussion about the use of the camera were contrary to company policy about nonessential conversation in the cockpit below 10,000 feet. This further illustrates the captain's disregard for standard operating proeedures, Such a subtle aspect cannot be directly attributed to the cause of this accident; however, it does illustrate an apparent lack of concern about the approach on the part of the captain. The Safety Board supports sterile cockpit procedures which exclude distractions during critical phases of flight. 2.4 survival Aspecis The crash forces which were transmitted to the occupants during the initial impact and subsequent ground slide were of insufficient magnitude to produce injuries. This is supported by the fact that three passengers were unrestrained by their seatbelts during most of the deceleration und they managed to escape uninjured. Because the evacuation was completed so quickly, no injuries or fatalities resulted from the fire. Based on statements of the flight attendants, some persons might have been trapped and killed by smoke and fire if the evacuation had taken omy @ few seconds longer. The loss of more than one-half the exits because of impact damage and fire made the timely evacuation all the more noteworthy. The fact that the aft arstair exit was not opened was nearly catastrophic because one flight attendant and som. passengers were almost trapped in that area. It coule not be determined if the pneumatic ~%4- emergency blow-down system would have forced the exit open; however, the fact that the fiight attendant did not know how to actuate the emergency system is a serious concern.

[ANALYSIS > THE ACCIDENT] It is recommended that the sterile cockpit light be turned on at 10,000 feet.” The "In Range" checklist contains the following: "Note: Captain will ascertain proper time to turn sterile cockpit light on." 2. ANALYSIS 2.1 The Accident The investigation revealed that the flightcrew was properly certificated and qualified to conduct the flight. The airsraft was properly certificated, equipped, and ~20- maintained. The landing gross weight was within limits for the reported winds and the 30° flap setting. The overtoud condition imposed on the right landing gear was caused by two conditions: the upslope of the area where the touchdown was made and the descent rate of the aircraft at touchdown. The investigation revealed that the shear load imparted to the landing gear as a result of the upsloping terrain was 824 ft/min (13.74 [t/sec), which would have been below the design strength if tne aircraft had beer on a level runway. Similarly, the calculated vertical descent rate (600 to 750 ft/min) would have imparted a shear load to the landing gear well below the design strength for a touchdown on a level runway. The combination of the two forees, howevcr, exceeded the design strength of the gear. Also, the right main landing gear sustained the full force of the impact without the left main landing gear sharing the load of a simultaneous contact. Therefore, the combination of the upslope at the touchdown point and the vertical descent of the aircraft caused the right main landing gear to separate. The Safety Board's analysis of the evidence in this accident focused on the reasons why the aireraft landed short of the rur.:‘ay. The investigation revealed no mechanical or meteorological reason which coud have caused the short landing. Examination of the wreckaze and a kinematic analysis of ithe dynamics of the touchdown revealed that the design strength of the right main lanc’ .¢ gear structure was exceeded by the forces of the impact. The right main landing gear separated as designed, precluding worse damage to the wing and fuselage structure and preventing a serious fuel spill at impact. The events subsequent to the initial touchdown were incidental only to the survival aspects of the accident. It is apparent from the statements of the four flightdeck occupants and crom the CVR and FDR information that the landing pattern at Yap was flown low and flat, which was not the standard prescribed procedure. Nevertheless, all four flightdeck oceupants believed that the aircraft was going to make a safe landing until tie aircraft was about 50 feet above the runway and the captain reduced the thrust to idle. Although the first officer, second officer, and the mechanic were concerned about the final approach being low, they apparently believed the aircraft would land on the runway until the power was reduced. The captain stated that he still believed that the aircraft would Jand on the runway, although closer to the threshold than he had planned. Airspeed was maintained at or near reference speed until the point where power was reduced about 50 feet above the runway. At that point, the descent rate increased rapidly when the thrust was reduced to idle.

[ANALYSIS > THE ACCIDENT] Even an "experienced" first officer could be reluctant to correct a captain. In this case, the first officer did advise the captain about being low; however, his similar lack of experience into Yap may have limited his ability to make a more definite evaluation and to recommend proper action. Even though he had recent experience as a B-727 captain and should have been aware of the proper procedure for flying such an approach, his position of first officer could have deterred him from taking more action in expressing bis concern about the approach. It is unlikely that even an "experienced" first officer could have prevented the captain from suddenly reducing the thrust to idle. There was insufficient time for the other members of the flighicrew to react and prevent the accident. Therefore, although the unwritten practice of providing an "experienced" first officer for newly trained captains in Air Micronesia operations may provide e higher level of safety, the existing training and experience requirements for air carrier operators should provide for safe operations even for a newly assigned flighterew. 2.3 Visual Ilusions and Distractions Another aspect in this case examined by the Safety Board was the possibility that the captain of Flight 614 was confused about the proper glidepath and touchdown point because of visual illusions. The heat waves he reported coming off the trees while on the final approach should not have presented a problem. The other crewmembers did not report such a phenomenon. If the aircraft had been on a proper 3° glidepath, the captain would not have experienced the condition. It certainly should not have caused Te Roe ~ ~ ~ ‘ - ~ 2 wy - : a J . . . = a's . . + wy a . we RR Pe Fare . - B ‘ . a . . ‘ . . a. Rel: ee f a . 6. te : . * - er 4 a Fs ~ . a + hw * , * - oe ie ~23- sufficient distortion of his view of the runway to cause him to aim short of the 1,U00-foot touchdown zone. Furthermore, at the point where he reduced power to idle, sucit conditions would have no longer existed and, therefore, should not have caused him to believe the runway was made. A second visual illusion aspect considered by the Safety Board was the possibility that the runway shape, including the undefined edges (see figure 1), may have contributed to the captain's faulty planning of his approach and landing flare. It is apparent that the runway appears wider in the first few hundred feet than its published 100-foot width. Similarly, it appears narrower for the remaining length because of the grass growing through the runway surface. The classic problem of runway width causing illusions pertains to the fact that the pilot us-s the apparent convergence angle of the runway edges in perspective to estimate length. Increasing or decreasing the distance between the lines can create illusions of shortening or lengthening of the pilot's perception of the runway length.

[ANALYSIS > THE ACCIDENT] If he had turned on the final approach at the same distance but at the proper altitude of 500 feet, he would have been on a normal 3° approach slope angle to the 1,000-foot aim point. However, the low base~-leg aititude and turn to the final approach required a flat approach slope angle of about 1.5° and a low rate of descent. He probably flew the a»proach in this manner to attempt a short field-type landing. Because he failed to establish a proper glidepath, his sight picture of the runway, as compared to a standard pattern, would have been abnormal, and more thrust would have been required to hold the lower-than-normal descent rate. This type of dragged-in, flat approach places an aircraft in a difficult situation with respect to windshear, downdrafts, or loss of thrust. Because the margins for error are much less in this type of approach, the FAA and airline companies prescribe standard stabilized approach procedures for jet transport category aircraft. A standard flight pattern procedure by the captain was all the more important in this ease because this was his first unsupervised landing at Yap since he resumed flying a B-72" aireraft. His recent requalification in the B-727 and limited familiarity with Yap should have alerted him to use the preseribed procedures. If he had, he would have had a greater margin for error. If he had reduced the throttles to idle at 50 feet over the runway surface during a prescribed approach, a hard landing probably would have resulted, but it is not likely the aircraft would have been damaged. Tre transition to a landing attitude begun at 5(}} feet from a normal 3° approach slope angle and the prescribed smooth thrust reduction will generally result in a normal lending, whereas a dragged-in, flat approach requires excess power. 2.2 Training Aspects The Safety Board believes that the captain's premature reduction of thrust on this final approach may have resulted from a habit pattern developed during nis previous experience in landing the DC-30. Specifically, the DC-10 has mass/energy and aerodynamic characteristics which produce a greater tendency to float in ground effect than does the B-727. Further, the DC-10 does not necessarily require comparatively es much thrust carried until at or near touchdown as does the B-727. Thus, the captain's prior experience in landing the DC-10 could have contributed to the development of & thrust reduction habit pattern which, although appropriate to the DC-10, was not appropriate for the B~727, especially during a low, flat approach in the B-727. The captain certainly should have been aware of the aircraft differences from his training; however, he did have a iong delay from his last B-727 training flight to his first line fight (61 days). He also returned to flying the DC-10 before his 8-727 line flying. This training sequence and time factor does occur in routine airline operations, especially following a reduction-in-force or other schedule changes.

[ANALYSIS > THE ACCIDENT] The captain certainly should have been aware of the aircraft differences from his training; however, he did have a iong delay from his last B-727 training flight to his first line fight (61 days). He also returned to flying the DC-10 before his 8-727 line flying. This training sequence and time factor does occur in routine airline operations, especially following a reduction-in-force or other schedule changes. St he te ein The proced :«s followed in this case meet a}}: the Federal seguiations and have not been shown to be improper in the past. Ideally, transition or requalification training should folicw 2 pattern whereby the pilot gces from one aircraft model to training in another and directly into line flying in the second. Practically, this situation is not always possible because of airline operstional and schedule requirements and hes not been identified as a factor in past uirline accidents. However, this situation must be vonsidered te be a factor in this accident, because if the captain had flown a proper pattern, this accident might not have occurred. The captain's statement that he had flown his training flights into Yap and Truk in a manner similar to the accident approach was not substantiated by the check captain. Moreover, examination of the FDR data for the captain's landing at Saipan on November 21, 1980, showed that he also flew a flat approach to that runway. He said he did so to remain clear of clouds, even though his final approach path was below the VASI plide slope. The interview with the check captain who gave the captain his line qualification for Air Micronesia confirmed that a 3° glide slope with about a 700 ft/min rate of descent js taught, even for Yap and Truk. He stated that he stressed the 1,000-foot aim point with thrust maintained te touchdown. The check captain stated that deviating beiow the VASI glide slope is not condoned, especially to avoid clouds, because the VASI is the aid most necessary to insure a proper glidepath and to prevent a short landing. The Safety Board could not determine a reason for the captain te ignore the training and procedures established for such landings. The company's unwritten practice of providing a first officer who was experienced in Air Micronesia operations for captains who were new to Air Micronesia operations was compromised when the scheduled first officer called in sick. Nevertheless, the captain's training and experience should have provided for a safe flight. Although an "experienced" first officer would be a plus for a "new" captain, in the ¢ se where a captain deviates from established procedures, even a highly experienced first officer may not be able to prevent an aceident. Even an "experienced" first officer could be reluctant to correct a captain. In this case, the first officer did advise the captain about being low; however, his similar lack of experience into Yap may have limited his ability to make a more definite evaluation and to recommend proper action.

[ANALYSIS] Section 2.3 discusses this anomaly and the role that it played in the flight crew’s inability to stop the airplane and avoid a runway overrun. 2.2.2.3 Captain Leadership and Decision-making The captain’s decision to land with a significant tailwind instead of a headwind is concerning as are the captain’s statements on the CVR that indicated uncertainty regarding the landing runway. Specifically, after the go-around decision, the captain stated that the airplane was going to make a 180° turn and come back for a landing. About 1 minute later, after hearing the first officer’s radio transmission (to the pilot of another airplane in the area) about coming back for the visual approach to runway 13, the captain stated, “three one” twice and “back door?” The first officer stated that he thought that the airplane would be landing on runway 13, to which the captain replied, “oh okay…sure…we’ll try again.” (It is possible that the first officer did not realize that the captain’s 180° turn remark indicated his intent to land on runway 31.) A few seconds later, the first officer asked if the captain was “okay” with a landing on runway 13, and the captain replied that he was thinking about landing on the opposite runway because of wind shear. Less than 1 minute later, the flight crew learned that the wind was from 300° at 24 knots. The first officer then asked the captain whether they should conduct another go around. After receiving no response to his question, the first officer stated, “we’ll try it again,” to which the captain replied, “last try.” These exchanges showed that the captain understood that runway 31 was the preferred runway for landing based on the wind information. However, the captain demonstrated poor CRM by not explicitly sharing his position about runway 31 until after the first officer told the other pilot operating in the area that the airplane would be landing on runway 13. Afterward, when the first officer stated to the captain that 1 minute after touchdown) indicated that a trace amount of precipitation had fallen since 1656, but the time period that the precipitation had accumulated since 1656 could not be determined. Aviation Accident Report NTSB/AAR-21-05 79 he thought the airplane would be landing on runway 13, the captain immediately stated, “oh okay…sure…we’ll try again.” The captain should have recognized that this situation called for a critical evaluation of the available data as part of the crew’s discussion of the issue. Unlike the go-around decision, the decision about which runway to return to for landing did not need to be made within seconds. Several PenAir pilots stated that, for a landing at DUT, they would choose the runway that favored the most recently reported wind. However, the captain did not conduct the landing on runway 31 and instead yielded twice to the first officer’s understanding, based on cursory discussions, that runway 13 should be used.

[ANALYSIS] However, before the second landing attempt, the airplane was operating in the airport traffic pattern and not from an approach, so the company’s stabilized approach criteria would be difficult to correlate to this portion of the flight. In addition, constraints associated with the airport (the surrounding terrain and the nonalignment of the instrument approaches with the runways) might have presented challenges to becoming established by 500 ft, even during routine approaches not involving the traffic pattern. (Because of these and other challenges associated with operating at DUT, PenAir designated DUT as an airport that required PICs to have company-specific experience and a corresponding qualification before the PICs could operate there, as discussed in section 2.4.1.) The airplane crossed the runway 13 threshold at an indicated airspeed of 127 knots and touched down at an indicated airspeed of about 125 knots and a groundspeed of about 142 knots. The runway was dry at the time of touchdown, as indicated by passenger videos of the touchdown and landing rollout as well as pavement images recorded by video from two off-airport security cameras.111 was evaluating a performance application for future in-flight use. (The program was not included in company manuals and was not approved by the FAA. The accident captain had the concurrence of the chief pilot to evaluate the application.) 110 As previously stated, the captain decided to land with 20° of flaps instead of 35° because he thought that the latter could lead to floating and difficulty pinpointing the landing spot. 111 The NOTAM included in the weather packet that the flight crew received before departure indicated that the runway was damp with no standing water. (The NOTAM was issued more than 9 hours before the accident airplane touched down.) The AWOS 1-minute observation at 1741 (about Aviation Accident Report NTSB/AAR-21-05 78 Less than 1 second after MLG touchdown, the captain reduced power to ground idle, and the flaps began to automatically retract from 20° to 15°. About 6 seconds later, the captain applied maximum reverse power. At that time, the airplane’s groundspeed was about 100 knots. After the first officer called out “brakes,” the captain replied, “I got em all the way.” However, the airplane was not decelerating as expected. A flight deck master caution annunciated multiple times during the landing rollout; these annunciations were related to the antiskid brake system anomaly. Section 2.3 discusses this anomaly and the role that it played in the flight crew’s inability to stop the airplane and avoid a runway overrun. 2.2.2.3 Captain Leadership and Decision-making The captain’s decision to land with a significant tailwind instead of a headwind is concerning as are the captain’s statements on the CVR that indicated uncertainty regarding the landing runway.

[ANALYSIS] The calculated stopping distances for a flaps 35 configuration were several hundred feet less than similar cases with a flaps 20 configuration. Aviation Accident Report NTSB/AAR-21-05 75 airplane was “goin’ around.” The flight crew then conducted a go-around and re-entered the VFR traffic pattern.105 The CVR also showed that the crewmembers did not adequately discuss how to manage the flightpath in the VFR traffic pattern. For example, the captain did not maintain the altitude setting, as evidenced by a flight deck altitude alert and the first officer’s statement noting that the airplane was 200 ft above the 1,000-ft setting. Also, after the first officer increased throttle power for the go-around, the captain instructed the first officer to “slow me up…back off on power.” The NTSB notes that, for a visual approach to runway 13 that results in a goaround, pilots fly along the back door path to Unalaska Bay and then re-enter the traffic pattern for either runway 13 or 31. Thus, after the accident go-around, the flight crew could have decided to re-enter the traffic pattern in the opposite direction for a landing on runway 31. However, the captain conducted the second landing attempt at DUT using the front door path to runway 13. 2.2.2.2 Second Approach and Landing While the airplane was on the base leg of the VFR traffic pattern for runway 13, the captain asked the first officer to request a wind check from the DUT weather observer, which he did about 1 minute 20 seconds before touchdown.106 The weather observer reported that the wind was from 300° at 24 knots; CVR statements indicated that the flight crew was surprised to learn about this significant tailwind for runway 13. The PenAir CFM, which was based on manufacturer information, indicated that the Saab 2000 had a 15-knot tailwind limitation. During postaccident interviews, the flight crewmembers stated that they were aware of the tailwind limitation but thought that the wind direction and speed did not warrant a change of the runway for landing. The flight crew’s continuation with the planned landing on runway 13 despite the knowledge of a tailwind that exceeded the manufacturer’s tailwind limitation was consistent with plan continuation bias, which is an unconscious cognitive bias to 105 At 1735:29, the DUT weather observer reported that the wind was from 300° at 8 knots. The flight crew did not request this wind report and likely did not pay attention to the report because it was transmitted when the flight crew would have been focused on the increased workload associated with the go-around. 106 About 1.5 minutes before the first officer’s wind check request, another pilot in the area was communicating with the first officer.

[ANALYSIS] About 1 minute later, after the first officer told a pilot in the area that the airplane was coming back around for the visual approach to runway 13, the captain stated “three one” twice. When the first officer stated that he thought that the airplane would be landing on runway 13, the captain responded, “oh okay…sure…we’ll try again.” Section 2.2.2.3 discusses this information further. 109 The captain stated, “unfactored landing distance would be five…three thousand fifty-eight feet,” but the NTSB notes that the captain had apparently misspoken when he stated “five” given that the runway was 3,900 ft. The 3,058-ft distance was likely from a performance application that the captain was testing on his iPad. The CVR appeared to have captured the captain demonstrating the application to the first officer earlier in the flight. Further, some of the company pilots who were interviewed as part of this investigation mentioned that the captain (as the PenAir director of training) Aviation Accident Report NTSB/AAR-21-05 77 configuration decided by the captain) and a 15-knot tailwind (the maximum tailwind value in the tables, consistent with the tailwind limit in the CFM), the performance binder tables indicated that the unfactored landing distance would be 3,536 ft.110 The landing distance available at DUT was 3,900 ft; thus, with a tailwind of 15 knots or less, sufficient landing distance should have been available for the flight. However, if the flight crewmembers had considered landing performance data based on the reported 24-knot tailwind, they could have realized that the tailwind exceeded the maximum tailwind in the performance binder tables and might have reconsidered the decision to land on runway 13. FDR data showed that the flight crew had not aligned the airplane with the runway heading until the airplane was below an altitude of 500 ft agl, which was inconsistent with the company’s stabilized approach guidance indicating that, for a visual approach, the airplane must be established on the correct course by 500 ft above the airport elevation. However, before the second landing attempt, the airplane was operating in the airport traffic pattern and not from an approach, so the company’s stabilized approach criteria would be difficult to correlate to this portion of the flight.

[ANALYSIS] 2. Analysis 2.1 Introduction This accident occurred after the flight crew landed the Saab 2000 airplane on runway 13 at DUT. The captain, who was the pilot flying, reported that the initial braking action was normal but that, as the airplane traveled down the runway, the airplane had “zero braking” despite his application of maximum brakes. The airplane subsequently overran the runway and came to rest beyond the airport property. The aircraft performance study for this accident found that the airplane should have had the landing performance capability to stop within the landing distance available on runway 13 or the RSA distance given the airplane’s energy state, MLG touchdown location, environmental conditions, and runway surface conditions. The following analysis summarizes the accident (section 2.2) and evaluates the following: • brake system issues (section 2.3), including the crossed wiring of the wheel speed transducer harnesses in the antiskid brake system; • operational factors (section 2.4), including PenAir’s PIC airport qualification policy; and • FAA oversight (section 2.5). After completing a comprehensive review of the circumstances that led to this accident, the investigation established that the following factors did not contribute to the cause of the accident: Flight crew qualifications. The flight crew was properly certificated and qualified in accordance with Part 121. Company qualification requirements are discussed in section 2.4.1. Flight crew medical conditions. The flight crew held valid and current medical certificates. A review of the flight crew’s work and sleep schedules and FAA medical certificate records showed no evidence of factors that could have adversely affected the performance of either crewmember on the day of the accident. The captain had previously been diagnosed with sleep apnea, but he used a continuous positive airway pressure machine consistently during sleep periods before the accident flight. Airplane conditions. The airplane was properly certificated, equipped, and maintained by PenAir in accordance with Part 121. No evidence indicated any structural or engine failures before the runway overrun occurred. Issues involving the antiskid brake system are discussed in section 2.3. Aviation Accident Report NTSB/AAR-21-05 72 Thus, the NTSB concludes that none of the following were factors in this accident: (1) flight crew qualifications and airplane certification, which were in accordance with US regulations; (2) flight crew medical conditions; and (3) the airworthiness of the airplane’s structures and engines. 2.2 Accident Summary 2.2.1 Preflight Planning PenAir used a wind chart for certain airports, including DUT, so that the dispatcher and the captain of a flight could evaluate the wind conditions at the destination airport.

[ANALYSIS] Unlike the go-around decision, the decision about which runway to return to for landing did not need to be made within seconds. Several PenAir pilots stated that, for a landing at DUT, they would choose the runway that favored the most recently reported wind. However, the captain did not conduct the landing on runway 31 and instead yielded twice to the first officer’s understanding, based on cursory discussions, that runway 13 should be used. The NTSB concludes that the captain demonstrated inadequate aeronautical decision-making skills regarding which runway to use for landing and a lack of flight deck leadership by continuing the landing to a runway with a significant tailwind. 2.2.3 Evacuation and Emergency Response The accident was survivable for all the airplane occupants except for the passenger in seat 4A, who was struck and killed by a propeller blade from the left engine that separated and entered the cabin during the impact sequence. The serious injury to the passenger in seat 5D occurred when he was struck by debris of an unknown origin. Most of the eight passengers who sustained minor injuries received those injuries from slips and falls during the evacuation, primarily from the right wing, which was wet because of rain. The evacuation resulting from the accident was unplanned. PenAir’s flight attendant manual indicated that the flight attendant was “solely responsible for the organization and evacuation of the passengers.” The manual also stated that evacuations should begin “as soon as the airplane stopped completely” and that no signal from the captain was necessary for an unplanned emergency situation. The flight attendant assessed the conditions outside of passenger windows on the left side of the airplane and appropriately determined that her exit (the main cabin door on the left side of the airplane) was unusable. She immediately ordered an evacuation through the right-side overwing exit and the right service door in the aft cabin, which were the only exits used during the evacuation. Passengers evacuating from the right service door faced a significant drop to the ground due to the airplane’s unusual, nose-down attitude and lack of an escape means. The passenger who opened the right service door stayed beneath the door after evacuating and assisted other passengers to the ground. Aviation Accident Report NTSB/AAR-21-05 80 Passengers from row 5 and aft evacuated immediately, but the evacuation of airplane occupants from row 4 and forward (including the three crewmembers) was delayed because of the critical injuries to the passenger in seat 4A. This passenger was removed from his seat by other passengers and was placed in the aisle immediately after the airplane stopped (which effectively blocked the egress path for those occupants in row 4 and forward) so that other passengers could attend to his injuries until emergency responders arrived. After most of the passengers from and aft of row 5 had evacuated, a local doctor (who witnessed the accident) boarded the airplane and began providing medical assistance to the critically injured passenger, and emergency medical technicians boarded the airplane and removed this passenger. The remaining airplane occupants exited the airplane after the critically injured passenger was removed, and all were off the airplane about 19 minutes after the accident.

[ANALYSIS] The flight crew did not request this wind report and likely did not pay attention to the report because it was transmitted when the flight crew would have been focused on the increased workload associated with the go-around. 106 About 1.5 minutes before the first officer’s wind check request, another pilot in the area was communicating with the first officer. At that time, the weather observer reported that the wind was from 290° at 16 knots with gusts to 30 knots. The CVR recording showed that the flight crew did not discuss this wind information. Aviation Accident Report NTSB/AAR-21-05 76 continue with an original plan despite changing conditions. With plan continuation bias, the closer a pilot gets to a destination, the stronger the bias becomes (Woods 2020).107 After receiving the weather report indicating a 24-knot tailwind, the first officer asked the captain, “do you wanna…back out…do it again?” With no definitive response from the captain, the first officer stated, about 10 seconds later, “we’ll try it again.” The captain then responded, “last try,” even though he had earlier commented about a possible landing on runway 31.108 Thus, the NTSB concludes that the flight crew’s decision to land on a runway with a reported tailwind that exceeded the airplane manufacturer’s limit was intentional, inappropriate, and indicative of plan continuation bias. PenAir’s Saab 2000 Performance Binder contained unfactored landing distance tables that listed the distances for various landing weights, flap settings, and wind conditions, and several company pilots stated that they would check landing performance data before an approach and after receiving the most recent weather information. PenAir procedures did not require pilots to conduct a time-of-arrival landing assessment, and no evidence indicated that the accident flight crew used the performance binder tables at that point to verify the landing distance required based on the tailwind and airplane flap setting. Earlier in the flight (while the airplane was en route to DUT), the CVR recorded the captain stating that the unfactored landing distance would be 3,058 ft, but that distance did not match any information in the Saab 2000 Performance Binder tables.109 At the airplane’s planned landing weight and with 20° of flaps (the 107 Plan continuation bias is known to negatively affect aeronautical decision-making; as a result, the subject is addressed in FAA guidance, training, and testing materials for pilots. 108 For example, about 50 seconds after the go-around, the captain stated that the airplane would go back out, do a 180° turn, and come back in, which indicated that the airplane would be returning to land on runway 31. About 1 minute later, after the first officer told a pilot in the area that the airplane was coming back around for the visual approach to runway 13, the captain stated “three one” twice.

[ANALYSIS] None of the airplane occupants were injured, and the airplane sustained minor damage. The NTSB’s investigation found that, during the landing rollout, the first officer felt the antiskid system releasing, so the captain took control of the airplane. The captain felt initial braking action, but all braking action was then lost. Postaccident examination of the airplane found that, among other brake system anomalies, the wiring to the left inboard and left outboard wheel speed transducers was crossed.118 This and the other brake system anomalies were not detected during antiskid system testing that was performed about 2 weeks before the incident. The NTSB determined that the probable cause of this incident was maintenance personnel’s “inadequate inspection of the aircraft…in that they did not properly diagnose discrepancies in the antiskid braking system.”119 The second incident occurred on October 9, 2007, when a United Airlines Airbus A320, N431UA, departed the runway and impacted runway lighting during landing at O’Hare International Airport, Chicago, Illinois. All 127 airplane occupants were uninjured except for 1 flight attendant and 1 passenger, who sustained minor injuries. The airplane sustained minor damage. The NTSB’s investigation found that, during the landing, the left MLG inboard wheel went to a high braking level and that the left MLG outboard wheel did not apply braking. The NTSB also found that the wiring for the airplane’s left MLG inboard and outboard antiskid tachometer (Airbus’ nomenclature for a wheel speed transducer) was reversed during scheduled 117 As previously stated, one of the multiple analyses conducted as part of system safety assessments is a functional hazard assessment, which is performed to determine potential system failures that could lead to hazards that could affect the airplane and its occupants. 118 The NTSB’s examination of the airplane also found that the left inboard antiskid valve brake and return lines were crossed and that the left inboard skid detect circuit was inoperative. In addition, the left MLG outboard tire was “blown.” 119 For more information about this incident, see case number ATL95IA043 at the NTSB’s website. Aviation Accident Report NTSB/AAR-21-05 88 maintenance. The NTSB determined that the probable cause of the incident was “the misrouted and reversed antiskid wiring by vendor maintenance personnel leading to the runway excursion.” Contributing factors to the incident included the operator’s maintenance procedures for the dual tachometer replacement, which were unclear to the vendor’s maintenance personnel.120 The third incident occurred on February 25, 2008, when a United Airlines Airbus A320, N442UA, departed the right side of the runway during landing at Jackson Hole Airport, Jackson, Wyoming. All 125 airplane occupants were uninjured except for 1 passenger, who received minor injuries during the evacuation. The airplane sustained minor damage.

[ANALYSIS] After most of the passengers from and aft of row 5 had evacuated, a local doctor (who witnessed the accident) boarded the airplane and began providing medical assistance to the critically injured passenger, and emergency medical technicians boarded the airplane and removed this passenger. The remaining airplane occupants exited the airplane after the critically injured passenger was removed, and all were off the airplane about 19 minutes after the accident. Although the evacuation delay for some occupants was not ideal, the flight attendant used her training and judgment to suspend the evacuation based on the need for the critically injured passenger to receive emergency medical attention. This decision was understandable and was supported by the flight crewmembers once they entered the cabin.112 In addition, the lack of a usable exit at the front of the airplane and the cramped conditions of the small airplane cabin could not be overcome. The emergency response to the accident included the on-airport ARFF vehicle as well as personnel from the Unalaska Fire Department and Unalaska Police Department, who responded to the accident site based on the mutual aid agreement in DUT’s Airport Certification Manual. Review of the video files and computer-aided dispatch reports indicated that the airport’s ARFF vehicle arrived on scene about 2 minutes 13 seconds after the airplane came to a stop. The first off-airport emergency vehicle arrived about 3 minutes later, and the critically injured passenger was removed from the airplane about 9 minutes afterward. The NTSB concludes that the evacuation delay for the crewmembers and some passengers was reasonable given the need to provide emergency medical attention to the critically injured passenger, and the emergency response was timely and effective. 112 During a postaccident interview, the captain stated that “there was no immediate danger to anyone in that aircraft at that point in time” and that the critically injured passenger “was the priority.” Aviation Accident Report NTSB/AAR-21-05 81 2.3 Brake System 2.3.1 Left Main Landing Gear Outboard Tire Because the outboard and inboard brakes transmit braking force to the wheelmounted tires, the tires need to be kept in a serviceable condition to ensure proper stopping performance. During the preflight inspection of the airplane, the first officer noted a flat spot on the left MLG outboard tire. The first officer took a photograph of the flat spot area and showed the photograph to the captain, who indicated that the flat spot was not a concern because the tire “wasn’t showing cord” (that is, no reinforcing plies were observed). PenAir did not require the flight crew to take any further action regarding this matter. The NTSB’s review of the first officer’s photograph found that no plies were showing on the tire, but the review also found the absence of grooves in the area of the flat spot. The Saab 2000 Aircraft Maintenance Manual stated that the tires were to be examined “for signs of damage, wear, and other…deterioration” and that a tire was to be removed “when tread has worn to the base of any groove.” According to this guidance, the tire should have been replaced before the accident flight.

[ANALYSIS] The captain decided to continue the approach to a landing because he and the first officer had determined that the landing criteria had been met, including the landing distance performance requirements for good braking action. The captain called the runway in sight as the airplane descended through an altitude of about 233 feet agl. Besides company and ATIS reports indicating that the runway was plowed, sanded, and chemically treated and was wet with snow, the flight crew had overheard ATC stating, as late as 1040, that arriving airplanes were being held for runway cleanup. The first officer stated that these communications “painted a picture” of what he and the captain could expect to see once the airplane broke out of the clouds—a runway that had at least some patches of runway surface visible. However, the captain and the first officer reported that the runway appeared white rather than dark or patchy, which was not 85 ATIS information Quebec (which was current at the time) reported the visibility as 1/4 mile, which was below the 1/2-mile visibility required for the approach. However, the approach could still be performed because the RVR was above 2,400 feet. 86 Before final approach, the flight crew had not discussed the possible effects that a tailwind could have on landing performance. One explanation for that could be that, at the time that the captain conducted the approach briefing, the reported wind (from ATIS information Papa) was from 040º at 7 knots, resulting in a direct crosswind (90º) for runway 13 with no tailwind component. Another explanation could be that Delta’s training program did not include instruction on landing with a tailwind and that the company’s MD-88 adverse weather operating procedures did not provide guidance for landing with a tailwind on a contaminated runway. NTSB Aircraft Accident Report 51 consistent with their expectations regarding the nature and extent of the contamination on the runway given the recent snow cleaning operations and the reports of good braking action.87 Delta’s MD-88/90 operating procedures for adverse weather indicated that, “when there is contamination on the runway…captains must evaluate crew, aircraft, and environmental conditions in determining the safety of operating their flight.” Key environmental conditions included braking action reports, the nature of contamination (water, wet snow, or dry snow), and the depth of the contamination. The flight crew knew that two preceding airplanes had reported good braking action on the runway. However, it would have been difficult for the crew to visually assess the nature and depth of the snow on the runway.88 Also, little time was available for the flight crew to reevaluate the decision to continue. Only 13 seconds had elapsed between the time that the captain called the runway in sight and the time of the 50-foot automated callout, when the captain would have been preparing to flare the airplane.

[ANALYSIS] However, it would have been difficult for the crew to visually assess the nature and depth of the snow on the runway.88 Also, little time was available for the flight crew to reevaluate the decision to continue. Only 13 seconds had elapsed between the time that the captain called the runway in sight and the time of the 50-foot automated callout, when the captain would have been preparing to flare the airplane. During these 13 seconds, the captain was engaged in several tasks, including adjusting the aim point to land closer to the approach end of the runway, decrabbing the airplane to align its longitudinal axis with the runway, and making aileron inputs to correct for drift. The NTSB concludes that, even though the flight crewmembers’ observations of snow on the runway were inconsistent with the expectations that they formed based on the field condition information that they received, their decision to continue the approach was not inappropriate because the landing criteria had been met. 2.1.5 The Landing The airplane touched down about 17 seconds after the captain called the runway in sight. During the 27 minutes between the time that runway 13 was last cleared of snow and the landing, snow continued to fall, with fresh snow covering the runway. As previously stated, the 0903 NOTAM, which was current at the time of the accident, reported 1/4 inch of wet snow on runway 13. The NOTAM was issued more than 1.5 hours before runway 13 was last cleared of snow (as discussed further in section 2.6.2); however, given the amount of time between the last runway clearing and the landing and the likely amount of snow that had accumulated during that time, the NOTAM likely described the approximate depth of the contamination on the runway surface at the time of the airplane’s landing. The accident airplane was the fifth arrival on runway 13 after it was last cleared. During this time, one report of medium braking action at the touchdown zone was received about 87 Although the flight crewmembers reported, after the accident, that the appearance of the runway was not what they expected based on the field condition information, the CVR showed that they did not consider that the snow that was falling could alter the landing performance calculations or that the braking conditions might have deteriorated from the good braking action reported by the flights crews of two airplanes that had recently landed on runway 13. However, the flight crew was not required to perform a go-around maneuver because the landing criteria were met and, as shown by FDR and radar data, the captain flew a stabilized approach. 88 Company landing criteria indicated that landing was not permitted with a snow depth of 1 inch or greater. Given the recent reports of good braking action on runway 13, the flight crew had no information that would have specifically prohibited the landing. NTSB Aircraft Accident Report 52 19 minutes before the accident landing, and reports of good braking action were received about 16 and 8 minutes before the accident landing.89 Because braking action reports are inherently subjective, the NTSB considered the results of the simulations it conducted with Boeing as part of this accident investigation.

[ANALYSIS] Because braking action reports are subjective, the captain attempted to assess the reliability of the reports indicating good braking action.84 At 1055:34, he asked the first officer, “wonder who reported braking action good? That’s another concern of mine,” to which the first officer replied, “it was United.” During a postaccident interview, the captain stated that he had confidence in this and the subsequent report of good braking action, which was likely because the reports were made by flight crews of other passenger jets operated by Part 121 air carriers. 82 According to the runway condition/braking action chart in Delta’s Operational Data Manual, a wet runway was expected to have good braking action. 83 During a postaccident interview, the LGA snow coordinator stated that the snow crews were not applying chemical to the runways at that point because snow had begun to accumulate and any application of chemical would have been immediately broomed, plowed, or blown off, reducing the chemical’s effectiveness. 84 SAFO 06012 indicated that a “reliable” braking action report would be “submitted from a turbojet airplane with landing performance capabilities similar to those of the airplane being operated.” NTSB Aircraft Accident Report 50 The flight crewmembers knew that, during a winter storm, changes to wind, visibility, or braking action could affect landing performance. They discussed that moderate snowfall could reduce visibility at the airport. They closely examined company policies for landing on contaminated runways and understood that a change in runway conditions from accumulating snowfall could increase the landing distance and that a change in wind could cause the flight to exceed crosswind limits. The NTSB concludes that the flight crew was well prepared for the approach and established landing requirements that were consistent with company policies. 2.1.4 The Approach At 1058:41, the controller cleared the flight for the ILS approach to runway 13. Shortly afterward, the controller told the flight crew that the RVR for runway 13 was greater than 6,000 feet, with a rollout RVR of 4,000 feet.85 At 1059:54 (about 2.5 minutes before landing), the captain told the first officer to ask the controller for a wind check because of the captain’s concerns regarding a “pretty good tailwind” on approach.86 The controller reported that the wind was from 020º at 10 knots. At that point, the tailwind component was about 4 knots, which was less than Delta’s 10-knot limit, and the crosswind component was about 9 knots, which was less than Delta’s guidelines for medium/poor runway conditions. The captain decided to continue the approach to a landing because he and the first officer had determined that the landing criteria had been met, including the landing distance performance requirements for good braking action. The captain called the runway in sight as the airplane descended through an altitude of about 233 feet agl.

[ANALYSIS] Given the recent reports of good braking action on runway 13, the flight crew had no information that would have specifically prohibited the landing. NTSB Aircraft Accident Report 52 19 minutes before the accident landing, and reports of good braking action were received about 16 and 8 minutes before the accident landing.89 Because braking action reports are inherently subjective, the NTSB considered the results of the simulations it conducted with Boeing as part of this accident investigation. The simulations showed that the wheel braking coefficient for the accident airplane was, at a minimum, consistent with the description of medium braking action in AC 25-32 and that the Delta MD-88 that landed uneventfully on runway 13 just before the accident landing had a similar wheel braking coefficient.90 In addition, the flight crew of the MD-88 airplane that landed on runway 13 just before the accident landing did not report any adverse landing conditions. Thus, the NTSB concludes that, although the runway was contaminated with snow, runway friction when the accident airplane landed was sufficient for stopping on the available runway length. For several seconds after touchdown, the airplane’s heading tracked with the runway centerline, indicating that the captain’s initial control inputs were sufficient to maintain directional control of the airplane. However, a combination of the 9-knot left crosswind component and asymmetric reverse thrust inputs (described below), as well as possible differences in runway friction for the left and right main landing gear, induced a left yawing moment after the airplane was on the runway.91 Under normal circumstances, this left yawing moment would be controllable through the use of right rudder pedal inputs, but, about 6 seconds after touchdown, the airplane’s heading began to deviate to the left from 131º, reaching 114º (20º off the runway’s magnetic heading of 134º) 6 seconds later. Reverse thrust is one method for decelerating the airplane on slippery runways, and it is most effective at high speeds. Delta’s Flight Crew Training Manual instructed MD-88 pilots to “activate reverse thrust with as little time delay as possible.” However, the manual also indicated that, when landing on a contaminated runway, MD-88 pilots should consider delaying thrust reverser deployment until nosewheel touchdown so that directional control would not be affected by asymmetric deployment. The captain deployed the thrust reversers almost simultaneously with the main landing gear touchdown. He reported that he had moved the thrust reverse levers back “one knob width on the reverser handle” to achieve Delta’s recommended target of 1.3 EPR on contaminated 89 As previously stated, the flight crew that reported medium braking action at the touchdown zone also reported poor braking conditions farther down the runway.

[CONCLUSIONS > FINDINGS] A callout when reverse thrust exceeds 1.3 engine pressure ratio during landings on contaminated runways could help avoid rudder blanking and a subsequent loss of directional control. 11. An automated alert could help minimize the possibility of reverse thrust engine pressure ratio exceedances during the landing rollout. NTSB Aircraft Accident Report 76 12. This accident demonstrates the continuing need for objective, real-time, airplane-derived data about runway braking ability for flight crews preparing to land with runway surface conditions that are worse than bare and dry. 13. The flight and cabin crews did not conduct a timely or an effective evacuation because of the flight crew’s lack of assertiveness, prompt decision-making, and communication and the flight attendants’ failure to follow standard procedures once the captain commanded the evacuation. 14. The flight attendants were not adequately trained for an emergency or unusual event that involved a loss of communications after landing, and the flight attendants’ decision to leave their assigned exits unattended after the airplane came to a stop resulted in reduced readiness for an evacuation. 15. This and other accidents demonstrate the need for improved decision-making and performance by flight and cabin crews when faced with an unplanned evacuation. 16. Aircraft rescue and firefighting personnel would likely have arrived at the accident scene sooner if they had received more timely and precise information about the accident and its location. 17. Even though the initial uncertainty regarding the total number of passengers aboard the accident flight, including lap-held children, did not lead to any adverse outcomes, such uncertainty could be detrimental under other accident circumstances, especially if search and rescue efforts are needed. 18. By not using its continuous friction measuring equipment during winter operations, LaGuardia Airport did not take advantage of a tool that would allow the airport to objectively assess the effectiveness of snow removal operations on contaminated runways. 19. The Federal Aviation Administration’s airport winter operations safety guidance is not sufficiently clear about the timing and need for updated runway condition reports, which could result in flight crew uncertainty about possible runway contamination. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the captain’s inability to maintain directional control of the airplane due to his application of excessive reverse thrust, which degraded the effectiveness of the rudder in controlling the airplane’s heading. Contributing to the accident were the captain’s (1) situational stress resulting from his concern about stopping performance and (2) attentional limitations due to the high workload during the landing, which prevented him from immediately recognizing the use of excessive reverse thrust. NTSB Aircraft Accident Report 77

[ANALYSIS] The captain (and the first officer) then sensed an absence of deceleration from the autobrakes, which drew the captain’s attention back inside the airplane so that he could verify that the spoiler handle was in the deployed position (a necessary action for the autobrakes to engage).95 The captain stated that, after he verified that the spoilers had deployed, he assumed that the autobrakes were not gripping. The captain also stated that, although he was concerned about the braking action, he had confidence in the ability of the autobrakes to achieve maximum braking for the available friction, so he did not immediately switch to manual braking. The captain further stated that he felt the airplane continue to slide to the left and that his efforts to counteract the slide with the rudder pedals were the primary focus of his attention at that point. However, the captain reported that the airplane did not respond to his efforts to steer back to the right. As the captain applied right rudder, the first officer made the “two in reverse” callout. The airplane’s heading then began to move to the left in response to the yawing moment. About 1.5 seconds after the first officer made the 110-knot callout and in response to the airplane’s change in heading, the first officer began calling for the captain to take the thrust levers out of reverse, stating (over a 1.5-second period) “out of reverse,” “come out of reverse,” and then in a louder voice “come out of reverse.”96 Although the first officer had not seen or heard indications of high reverse thrust, he suspected that rudder blanking (due to excessive EPR and the subsequent loss of rudder effectiveness) was the reason that the captain was unable to counteract 94 A standard runway safety area is 1,000 feet long by 500 feet wide, but some airports, including LGA, have nonstandard runway safety areas due to geographic or other limitations. The runway safety area at the approach end of runway 13 is 90 feet long and 150 feet wide. The runway safety area at the approach end of runway 31 (the departure end of runway 13) is 460 feet long by 500 feet wide. An engineered materials arresting system is in place at the departure end of runway 13. 95 As previously indicated, FDR data showed that the autobrakes activated almost simultaneously with the increase in reverse thrust. FDR data also showed that spoiler deployment occurred at the same time as thrust reverser deployment. The first officer indicated that he manually activated the ground spoilers because they did not automatically activate immediately. A postaccident examination of the spoiler system found no anomalies; thus, during the initial touchdown, the autospoiler actuator most likely did not activate immediately upon landing because the main landing gear wheels had not yet spun up to the necessary rotation speed. Less than 1 second after manually activating the spoilers, the first officer made the “spoilers up” callout. 96 At the time of the first officer’s “out of reverse” announcement, the left engine had reached 1.9 EPR, and the right engine had reached 1.8 EPR.

[ANALYSIS] However, the NTSB’s aircraft performance study showed that medium braking action conditions were present when the previous Delta MD-88 successfully stopped on runway 13. NTSB Aircraft Accident Report 49 to the first officer that company dispatch had not alerted them about the possibility of holding. The captain then asked the first officer if he had received any braking action reports from the controller. After the first officer replied that he had not yet received any reports, the captain repeated his frustration about dispatch’s lack of advanced notice about holding and expressed further frustration about the lack of timely braking action reports from ATC and dispatch (before the runway was closed for snow removal operations). The NTSB concludes that the flight crewmembers’ uncertainty about the runway conditions at LGA led to some situational stress for the captain. At 1045:38, the accident flight crewmembers heard a controller at the New York TRACON advise a flight crew of a poor braking action report for runway 13. The captain told the first officer “we can’t land with poor.” Shortly afterward, the controller advised another flight crew that an Airbus airplane had reported good braking action on runway 13. Afterward, the flight crew of a regional jet also reported that the braking action on runway 13 was good. During postaccident interviews, the accident captain recalled thinking that two reports indicating good braking action meant that conditions were adequate to proceed with the approach, and the accident first officer recalled thinking that two consecutive reports of good braking action were a “green light” to continue. ATIS information Quebec (issued at 1051) indicated that 1/4 inch of wet snow was observed on the runways at 0904 and that “all runways are wet and have been sanded and deiced with solid chemical.”82 However, the most recent NOTAMs at the time (issued at 0902 and 0903) reported that 1/4 inch of wet snow was on the runways but did not indicate that the runways had been treated. Further, according to radio communications between ATC and the LGA snow coordinator, about 1038 the snow coordinator reported to the local controller that “the runways have not been treated” and that “we’re just brooming and plowing.”83 (The flight crew did not know the information reported by the snow coordinator.) Thus, the ATIS report that was current at the time of the accident (Quebec), as well as the four previous ATIS reports (Lima, Mike, Oscar, and Papa, issued between 0751 and 1024), contained outdated and contradictory field condition information about the status of LGA’s runways. Because braking action reports are subjective, the captain attempted to assess the reliability of the reports indicating good braking action.84 At 1055:34, he asked the first officer, “wonder who reported braking action good?

[ANALYSIS] At 1104:00, the snow coordinator notified the local controller that runway 13 was closed and then proceeded onto the runway in his vehicle. Because the controller had not responded to the snow coordinator’s notification, he repeated that the runway was closed. At 1104:16, the local controller asked the snow coordinator whether the runway was closed, and the snow coordinator confirmed that information. About 14 seconds later (and 33 seconds after the initial notification that the runway was closed), the controller instructed the flight crew of Delta flight 1999 (the next arriving airplane for runway 13) to go around. On the basis of airport protocol and the letter of agreement with the ATC tower, the snow coordinator had likely assumed that the controller would have immediately closed the runway. However, this assumption led to a situation in which the snow coordinator’s vehicle was on the runway for about 23 seconds while Delta flight 1999 was on final approach. At the time of the controller’s go-around instruction, the flight 1999 airplane was only about 30 seconds from landing. The controller did not expect an abrupt closure of the runway and had likely assumed that the closure was due to the winter weather. (The snow coordinator did not state that the accident was the reason for the closure.) However, the controller had just attempted to contact the accident airplane’s flight crew six times without a response, so he might have reacted more quickly if he had integrated that information along with the runway closure information. Airport operations personnel notified the airport operations manager via cell phone that an airplane had departed the paved runway surface, hit a fence, and was leaking fuel. At the time, the airport operations manager happened to be engaged in a face-to-face conversation with the ARFF deputy chief and relayed the phone call information to him. At 1104:35, the deputy chief called the ARFF crew chief to prepare ARFF crews for a response based on the phone call information. However, because the airplane’s precise location was not provided as part of the NTSB Aircraft Accident Report 68 initial cell phone call, the only information that the ARFF crews had regarding the accident location was that the airplane had hit a fence. At 1104:38, a responding airport operations staff member notified the local controller about the accident. At 1104:48, another responding airport operations staff member told the controller, “you have an aircraft off [runway] 3-1 on the north vehicle service road. Please advise crash/rescue.” However, the EANS—the primary method for communicating an emergency at LGA—was not immediately activated by ATC tower personnel. The snow coordinator arrived on scene and told the controller at 1105:55 that the airplane was “leaking fuel on the left side of his aircraft heavily…his wing is ruptured.” Thirty seconds later (and about 3.5 minutes after the airplane came to a stop), ATC tower personnel activated the EANS, which reported “LaGuardia, Alert 3, all emergency vehicles respond.

[ANALYSIS] While in cruise flight at 0954:52, the flight crew consulted Delta’s contaminated runway crosswind limitations and recognized that the crosswind would be at Delta’s 10-knot limit if the runway braking action was reported to be medium/poor. The flight crew then requested, via ACARS, a field condition report from the dispatcher. The report indicated that braking action advisories were in effect but that no braking action reports were available. At 1005, the flight crew discussed the effect of moderate snow on visibility. Beginning at 1010:35, the crew discussed the landing distance using flaps 40, a 130,000-pound landing weight (which was less than the preflight planned landing weight due to fuel burn), and medium braking action. The flight crew recognized that the airplane would not be able to land with medium/fair braking action because the calculated landing distance for maximum autobrakes (7,800 feet) and maximum manual braking (7,200 feet) would exceed the available runway 13 length of 7,003 feet.81 As a result, the flight crew relayed information to the dispatcher indicating that the airplane could only land with good braking action. The flight crew also asked the dispatcher (at 1018) and the Washington ARTCC controller (at 1027) for braking action reports, but neither had any reports at the time to provide the crew because LGA operations personnel had been conducting snow removal operations on runway 13/31 (between 1006 and 1035) and no airplanes were landing. After the Washington ARTCC controller advised the flight crew that the airplane would be holding at the Robbinsville VOR because of runway cleanup, the captain expressed frustration 80 The dispatch release showed a planned fuel load (29,600 pounds) that was in excess of the required fuel load (about 26,000 pounds) to allow the flight crew to hold the airplane for an extended period or divert to one of the two designated alternate airports if necessary. However, the extra fuel would also increase the planned landing weight if holding or diverting was not necessary. 81 As previously stated, the landing performance data in Delta’s Operational Data Manual assumed a 1,500-foot air distance and a safety factor of 15%. These data are conservative to account for any uncertainty regarding the landing. As a result, the flight crew’s landing distance assessment showed that medium braking action would result in an insufficient stopping distance. However, the NTSB’s aircraft performance study showed that medium braking action conditions were present when the previous Delta MD-88 successfully stopped on runway 13. NTSB Aircraft Accident Report 49 to the first officer that company dispatch had not alerted them about the possibility of holding. The captain then asked the first officer if he had received any braking action reports from the controller.

[ANALYSIS] Operations on contaminated runways  MD-88 pilots received a flight safety electronic bulletin discussing contaminated runway awareness.  Delta was sponsoring two demonstration studies of aircraft-based technology for conducting runway friction assessments on contaminated runways. Special winter operations airport program  Delta identified LGA as a special winter operations airport for the 2015-2016 winter season. (As previously stated, LGA was not identified as a special winter operations airport for the 2014-2015 winter season, during which time the accident occurred.) NTSB Aircraft Accident Report 74  A Delta senior vice president sent a letter to Port Authority “urging” the adoption of a more robust friction assessment program.  Delta made “repeated requests” to LGA to issue hourly field condition reports, and LGA agreed (as discussed in section 2.6.2).  In December 2015, Delta’s flight safety department initiated a review of the special winter operations airport program to assess its effectiveness. The group conducting the review determined that the program was effective in identifying and mitigating potential runway excursion risks during winter operations. The group identified changes to enhance the program’s assessment of airports and the methods to mitigate identified hazards. Also, the group recommended several modifications to address the upcoming changes resulting from the TALPA program. NTSB Aircraft Accident Report 75 3. Conclusions 3.1 Findings 1. The flight crew was properly certificated and qualified in accordance with federal regulations and company requirements. Flight crew fatigue was likely not a factor in this accident. The airplane was properly certificated, equipped, and maintained in accordance with federal regulations. No evidence indicated any preimpact structural, engine, or system failures. 2. The flight crewmembers’ uncertainty about the runway conditions at LaGuardia Airport led to some situational stress for the captain. 3. The flight crew was well prepared for the approach and established landing requirements that were consistent with company policies. 4. Even though the flight crewmembers’ observations of snow on the runway were inconsistent with the expectations that they formed based on the field condition information that they received, their decision to continue the approach was not inappropriate because the landing criteria had been met. 5. Although the runway was contaminated with snow, runway friction when the accident airplane landed was sufficient for stopping on the available runway length. 6. The circumstances associated with the landing, including the snowier-than-expected runway, short runway length, and body of water off the departure end of the runway, likely exacerbated the captain’s situational stress and prompted him to make an aggressive input on the thrust reversers. 7. The captain was unable to maintain directional control of the airplane due to rudder blanking, which resulted from his application of excessive reverse thrust. 8. Even though the first officer promptly identified rudder blanking as a concern and the captain stowed the thrust reversers in response, the airplane’s departure from the left side of the runway could not be avoided because directional control was regained too late to be effective. 9. A solution to reliably limit reverse thrust engine pressure ratio values could benefit all pilots of MD-80 series airplanes. 10.

[ANALYSIS] However, during the accident landing, directional control of the airplane was affected by the disruption of airflow resulting from the high reverse thrust (which reached maximum values of 2.07 EPR on the left engine and 1.91 EPR on the right engine) and the runway friction, which, although sufficient, was limited. In addition, the higher reverse thrust on the left engine would tend to yaw the airplane to the left. Even though all Delta MD-88 pilots had been instructed to target 1.3 EPR on contaminated runways and practiced targeting that reverse thrust setting during simulator training, a habit of using more than 1.3 EPR might have prevailed when the captain was faced with the demands of the landing on the day of the accident. Situational factors associated with the landing also likely affected the timing and magnitude of the captain’s reverse thrust input. The captain knew that the conditions could be marginal for landing on runway 13 and that the landing would require most of the runway for the airplane to stop given the conditions and the airplane’s landing weight. In addition, the captain likely perceived the snowier-than-expected runway as a threat that, along with the relatively short runway, could impede his ability to stop the airplane within the available runway length. These factors, as well as runway 13’s nonstandard runway safety area and the presence of Flushing Bay directly off the departure end of the runway, likely compounded the captain’s previous stress resulting from the unavailability 92 The FDR did not record reverser handle position, so it is not possible to determine the captain’s actual input. During a postaccident interview, the captain stated that he would normally “tweak” the thrust reversers after his initial input to obtain the target EPR setting, but FDR data showed that he did not adjust the thrust reverse settings until the first officer’s callouts to come out of reverse. 93 The reviewed Delta MD-88 landing data showed that such asymmetry commonly occurred during the application of reverse thrust. NTSB Aircraft Accident Report 54 of braking action reports during the en route and descent phases of flight.94 The NTSB concludes that the circumstances associated with the landing, including the snowier-than-expected runway, short runway length, and body of water off the departure end of the runway, likely exacerbated the captain’s situational stress and prompted him to make an aggressive input on the thrust reversers. The captain recalled that, immediately after his application of reverse thrust, the airplane began sliding to the left, which caused him to focus his attention outside of the airplane and at the high snowbanks along the runway. The captain (and the first officer) then sensed an absence of deceleration from the autobrakes, which drew the captain’s attention back inside the airplane so that he could verify that the spoiler handle was in the deployed position (a necessary action for the autobrakes to engage).95 The captain stated that, after he verified that the spoilers had deployed, he assumed that the autobrakes were not gripping.

[ANALYSIS] The first officer reported needing 7.5 hours of sleep per night to feel rested, so he had a cumulative sleep deficit of about 2 hours (see table 3) during that period. Research indicated that such sleep deficits could result in very small performance-related effects (Belenky and others 2003). However, these minor effects would likely have been offset by circadian factors that promote alertness during the late morning hours (when the accident occurred). In addition, at the time of the accident, the flight crew had not been on duty excessively long (about 6 hours including the accident flight), flown multiple trip legs (the accident flight was the second leg of the day), or been awake for a long time (about 7.5 hours for the captain and 7 hours for the first officer). 78 Although no deficiencies were noted in Delta’s weather documentation, the company’s meteorology department did not anticipate the low IFR conditions or 1/4-mile visibility during the period and the heavy snow that occurred immediately after the accident. 79 ATIS information Mike reported 1/2-mile visibility with snow and freezing fog. Title 14 CFR 121.613 prohibits the dispatch of a flight operating under IFR unless appropriate weather reports or forecasts, or any combination thereof, indicate that the weather conditions will be at or above the authorized minimums at the estimated time of arrival. The minimum visibility for an instrument approach to runway 13 at LGA was 1/2 mile; thus, at the time of dispatch, the forecasted visibility for LGA was at the required minimum visibility. NTSB Aircraft Accident Report 48 the runways were wet and had been sanded and deiced with solid chemical, which was not current given the report of thin, wet snow in the 0738 NOTAM. The flight was dispatched according to 14 CFR 121.195.80 Dispatch calculations indicated that the landing distance would be 5,995 feet (based on a planned landing weight of 132,500 pounds, flaps 40, a 15% safety margin for a wet runway, and maximum braking). The operational landing distance assessment, which included the use of maximum autobrakes with good braking action and reverse thrust (among other factors), was 6,200 feet—about 800 feet less than the available length of runway 13. Thus, the information that the flight crew had at the time of departure showed that the airplane would be able to safely stop on the runway. 2.1.3 En Route Preparations and Decision-making After the airplane departed ATL (about 0924), the flight crew continued to monitor the weather conditions at LGA and assessed the factors that could affect stopping performance. While in cruise flight at 0954:52, the flight crew consulted Delta’s contaminated runway crosswind limitations and recognized that the crosswind would be at Delta’s 10-knot limit if the runway braking action was reported to be medium/poor.

[ANALYSIS] During a December 2015 meeting between LGA airport operations staff and the chief pilots for air carriers that operate at the airport, LGA staff indicated that it would update NOTAMs during winter operations each hour, regardless of whether runway conditions changed, and as necessary between hourly reports. The NTSB believes that this approach can help ensure that pilots receive timely and accurate information about the runway conditions at LGA. NTSB Aircraft Accident Report 73 2.7 Delta Air Lines Postaccident Actions According to Delta, as of June 7, 2016, the company had taken the following actions in response to the safety issues involved in this accident: MD-88 reverse thrust usage  On March 11, 2016, the normal reverse thrust for dry runways was adjusted to 1.3 EPR.  As of May 18, 2016, MD-88 landing distances were calculated using idle reverse thrust. – For dry runways, flight crews were to use 1.3 EPR. – For runways that are not dry, flight crews were to initially select idle reverse thrust. After reverse thrust symmetry was verified and the airplane was aligned with the runway track, crews could “methodically and gradually” increase reverse thrust to no more than 1.3 EPR if required to reduce stopping distance.  Delta’s technical operations department was implementing a new maintenance procedure to decrease asymmetric spool-up when reverse thrust is selected.  A safety management system safety risk assessment was initiated to review reverse thrust usage on the MD-88.  Company fleet bulletins and articles highlighted the use of reverse thrust on the MD-88. Operational landing distance data  On May 4, 2015, Delta activated the ACARS landing performance request tool for flight crew use. (Delta had begun testing the tool 2 months before the accident.) According to Delta, the landing performance request tool calculates runway-specific operational landing distance data and reduces the need for crews to reference the Operational Data Manual and interpolate and manually apply corrections.  Delta aligned its braking action table with the TALPA ARC recommendations. Operations on contaminated runways  MD-88 pilots received a flight safety electronic bulletin discussing contaminated runway awareness.  Delta was sponsoring two demonstration studies of aircraft-based technology for conducting runway friction assessments on contaminated runways.

[ANALYSIS] The FAA has consistently indicated that it would encourage, but not mandate, the use of airplane-based systems that determine runway surface conditions. However, given the success of CAST in effecting widespread voluntary introduction of safety improvements (including FAA-approved voluntary safety programs such as FOQA and the Aviation Safety Action Program), many operators might voluntarily equip their fleets with these systems if they are shown to be feasible, even without regulatory action by the FAA. Thus, if the FAA takes the 106 The NTSB understands that the FAA is still in the process of vetting and validating the calculations performed by this system and that work remains to determine how the calculated results are to be used by the company’s customers for operational decisions. NTSB Aircraft Accident Report 62 actions necessary to satisfy Safety Recommendations A-16-23 and -24, the intent of Safety Recommendation A-07-64 will also be satisfied without the FAA mandating the requested action. As a result, Safety Recommendation A-07-64 is reclassified “Closed—Acceptable Action/Superseded.” 2.4 Evacuation Issues 2.4.1 Evacuation Procedures After unexpectedly departing the paved runway surface, the airplane came to rest at a non-normal attitude with its nose over water and left wing damaged. The accident sequence also damaged the electrical systems, which resulted in the loss of the interphone and public address system as a means for communication. As a result, the two flight attendants in the aft cabin left their assigned locations and walked to the forward cabin (to obtain information from the flight crew and the lead flight attendant) while instructing passengers to stay seated and stay calm. The lead flight attendant left her assigned location to check on a passenger in the mid-cabin. However, the flight attendants were expected to stay near their assigned exit and remain in a state of readiness for an evacuation, which included assessing their exits to determine their status.107 Once the captain exited the flight deck, he met the flight attendants in the forward galley. The captain asked about the status of the forward and 2L door exits. The captain also asked whether the tailcone exit would also be available for an evacuation, and the flight attendant who had been seated at the tailcone replied “I don’t know.” Recognizing that no one was monitoring the aft cabin, one of the aft flight attendants walked back to her assigned position. A passenger stopped the flight attendant to tell her that a first responder was motioning to open the right overwing exits, to which she replied, “no, we need to wait until our captain instructs us to evacuate.” A first responder told the first officer, while he was in the cockpit, that everyone should evacuate the airplane via the right overwing exits due to fuel leaking from the left wing.

[ANALYSIS] The captain deployed the thrust reversers almost simultaneously with the main landing gear touchdown. He reported that he had moved the thrust reverse levers back “one knob width on the reverser handle” to achieve Delta’s recommended target of 1.3 EPR on contaminated 89 As previously stated, the flight crew that reported medium braking action at the touchdown zone also reported poor braking conditions farther down the runway. The accident flight crew was not aware of the medium braking action report from that flight crew but had overheard the controller at the New York TRACON advising another flight crew of the poor braking action conditions. 90 The airport RCAM as of June 2016 (see figure 5), which airports were not expected to use until October 2016, would have categorized the braking action on runway 13 at the time of the accident landing as medium (given the 1/4 inch of wet snow reported in the 0903 NOTAM). 91 According to the FAA’s Airplane Flying Handbook (FAA-H-8083-3A), during the roll after a crosswind landing, the wind acting on the airplane will have a “weathervaning” effect. The handbook also stated that “the greater the crosswind component, the more difficult it is to prevent weathervaning.” NTSB Aircraft Accident Report 53 runways.92 After this reverse thrust input, the EPR on both engines began to increase rapidly. FDR data showed that the left engine reverse thrust increased faster than that on the right engine (resulting in asymmetric reverse thrust).93 Further, the left engine reverse thrust increased from 1.3 to 1.6 EPR (the target setting for landing on dry runways) in 0.5 second. The NTSB’s review of data from 35 Delta MD-88 landings with at least one engine above 1.6 EPR showed that the average time between 1.3 and 1.6 EPR was about 1.25 seconds. Thus, the reverse thrust during the accident landing increased more rapidly than it did during the other landings, indicating that the captain’s input on the thrust reverser handles was relatively aggressive. It is possible that the captain had developed a habit of applying more reverse thrust than 1.3 EPR. Because most landings are performed on dry runways, Delta pilots, including the accident captain, were accustomed to using 1.6 EPR reverse thrust (or greater) after such landings. At landing and initial rollout speeds, the rudder should generate enough aerodynamic force to control the direction of the airplane. On a dry runway with an average wind, the rudder blanking that results from high reverse thrust is not significant because high frictional forces between the tires and the runway maintain the airplane’s direction during the landing roll. However, during the accident landing, directional control of the airplane was affected by the disruption of airflow resulting from the high reverse thrust (which reached maximum values of 2.07 EPR on the left engine and 1.91 EPR on the right engine) and the runway friction, which, although sufficient, was limited.

[ANALYSIS] During the accident landing, the runway was contaminated, and braking action advisories were in effect. In such situations, pilots might use higher amounts of reverse thrust because they are more concerned about their ability to stop an airplane within the available runway distance than they are about the risk of a loss of directional control. Also, pilots with a habit of routinely using more reverse thrust than needed might lack an awareness of the risks of EPR exceedances because such exceedances rarely result in any adverse outcomes. Another possibility for the exceedances of published EPR settings is that competing demands for the flight crews’ attention during landing might result in difficulties adhering to EPR targets despite efforts to do so. Landing is a high-workload phase of flight during which crews make required control inputs based largely on visual information acquired from inside and outside of the cockpit. After initially moving the thrust reverser handles, the PF must periodically examine the EPR gauges while steering the airplane down the runway and then adjust the reverser handles as needed in response to observed engine indications.101 Steering the airplane and monitoring the EPR gauges compete for the same visual and cognitive resources, which can 99 Boeing’s MD-80 Flight Crew Operating Manual, volume II, “Operating Procedures,” dated May 15, 2014, cautioned that 1.3 EPR reverse thrust was the maximum that should be applied on wet, slippery, or contaminated runways. The manual also indicated that 1.6 EPR reverse thrust was the maximum for dry runways except in an emergency. 100 Of the 11 MD-80 runway excursions (over a 20-year period) that are identified in table 7, reverse thrust exceeded 1.3 EPR during eight of the events, all of which involved runway contamination. Of those eight events, five involved reverse thrust at or above 1.6 EPR. 101 Because the thrust reverser handle position is not recorded by the FDR, the lag in engine spool-up time that occurs during actual operations is not known. The landing data reviewed as part of this investigation indicated that the engine spool-up time could be about 0.5 to 2 seconds. NTSB Aircraft Accident Report 57 cause the increased difficulty associated with one task to interfere with the performance of the other task. For example, the airplane steering task can be more difficult in snow conditions because of the challenges associated with discriminating key features of the runway environment. This situation would tend to increase the time allocated to acquiring information outside the cockpit and consequently degrade the effectiveness of the flight crew’s monitoring of indications (including EPR gauges) inside the cockpit.

[ANALYSIS] The tire marks over the next 200 feet of travel indicated that the airplane departed the runway at an angle of 10.4o to the left of the runway centerline. 40 2. ANALYSIS 2.1 General The flightcrew was properly certificated and qualified in accordance with applicable regulations and company requirements. All three crewmembers were experienced at their respective positions. Evidence from crew duty time, flight time, rest time, or off-duty activity patterns did not indicate that behavioral or physiological factors affected the flightcrew on the day of the accident. The ATC personnel involved with the flight were all properly certificated and qualified. The airplane was properly certificated, equipped, and maintained (with the exception of the FDR system) in accordance with FARs and approved regulations. The weight and balance were within allowable limits. In analyzing this accident, the Safety Board focused on flightcrew actions and decisions, B-747 procedures for slippery runway operations, the performance of air carrier training simulators for B-747 operations on slippery runways, flight attendant actions and cabin safety issues, Tower Air management oversight of maintenance and operations, FAA surveillance of Tower Air, and FAA policies and procedures regarding the evaluation of slippery runways. 2.2 Flightcrew Actions and Decisions 2.2.1 Pre-takeoff Events Although the flightcrew was not provided the runway friction values obtained by the airport operations crew, they had obtained sufficient indications from the slipperiness of the taxiways, the appearance of runway 4L, and the blowing snow to recognize that they were operating in a challenging environment of wind, reduced visibility, and runway slipperiness. Based on the existing surface and wind conditions on the day of the accident, the captain might have considered using runway 31L (which was more favorably oriented to the wind) for his departure. However, when the captain overheard the response of JFK ground control to another flight’s inquiry about runway 31L that it would remain closed for another couple of hours, he determined that runway 31L was not a viable option for departure. Although 5 minutes before the accident ATC changed the departure runway to 31L for traffic following flight 41, the Safety Board recognizes that the captain’s decision to use runway 4L was based on the limited information available to him at the time. Further, air traffic controllers were not required to offer flight 41 the option of switching to runway 31L, once the airplane was established holding short at runway 4L.

[ANALYSIS] Further, there are no means to compare measurement standards or translate the data into aircraft performance. A key issue is that no significant progress has been made in correlating stopping distance data from airplane manufacturers’ flight tests and calculations with the friction values obtained from measuring devices. An outcome of these correlations could be the establishment of objective standards for air carrier operations on slippery runways, perhaps extending to the establishment of appropriate minimum runway friction levels for operational use. The Safety Board concludes that the circumstances of this accident indicate that the issue of correlating airplane stopping performance with runway friction measurements should be revisited by the Government and the air transportation industry. Consequently, the Safety Board believes that the FAA should require the appropriate Aviation Rulemaking and Advisory Committee to establish runway friction measurements that are operationally meaningful to pilots and air carriers for their slippery runway operations (including a table correlating friction values measured by various types of industry equipment), and minimum coefficient of friction levels for specific airplane types below which airplane operations will be suspended. 55 3. CONCLUSIONS 3.1 Findings 1. The flightcrew was properly certificated and qualified in accordance with applicable regulations and company requirements. 2. The air traffic control personnel involved with the flight were all properly certificated and qualified. 3. The airplane was properly certificated, equipped, and maintained (with the exception of the flight data recorder system) in accordance with approved regulations. The weight and balance were within allowable limits. 4. The captain’s decision to attempt the takeoff on runway 4L was appropriate. 5. Asymmetric thrust was not a factor in the loss of directional control. 6. The captain’s failure to correct the airplane’s deviation from the centerline resulted from his overcontrolling the nosewheel steering through the tiller. 7. The captain of flight 41 first relied on right tiller inputs as the airplane continued to veer left, then applied insufficient or untimely right rudder inputs to effect a recovery. 8. Current Boeing 747 operating procedures provide inadequate guidance to flightcrews regarding the potential for loss of directional control at low speeds on slippery runways with the use of the tiller. 9. The procedural change by Tower Air to reevaluate and eliminate its standard procedure of guarding the tiller during the takeoff roll through 80 knots will make overcontrol of the tiller less likely for its own operations; however, other air carrier operators of the Boeing 747 may need to make similar changes to their procedures. 10. Current Boeing 747 flight manual guidance is inadequate about when a pilot should reject a takeoff following some indication of a lack of directional control response. 11. Improvements in the slippery runway handling fidelity of flight simulators used for Boeing 747 pilot training are both needed and feasible. 12. The captain’s failure to reject the takeoff in a timely manner was causal to the accident. 13. The inadequate Boeing 747 slippery runway operating procedures developed by Tower Air and the Boeing Commercial Airplane Group, and the inadequate fidelity of B-747 56 flight training simulators for slippery runway operations, contributed to the cause of this accident. 14.

[ANALYSIS] Although 5 minutes before the accident ATC changed the departure runway to 31L for traffic following flight 41, the Safety Board recognizes that the captain’s decision to use runway 4L was based on the limited information available to him at the time. Further, air traffic controllers were not required to offer flight 41 the option of switching to runway 31L, once the airplane was established holding short at runway 4L. Based on the absence of definitive runway friction measurements for runway 4L, reported winds of less than 15 knots (the maximum recommended crosswind component for B-747 takeoffs on slippery runways), the flightcrew’s reports of acceptable visibility down the runway, and the reported unavailability of the alternative runway 41 31L, the Safety Board concludes that the captain’s decision to attempt the takeoff on runway 4L was appropriate. 2.2.2 The Attempted Takeoff and Loss of Control Flight 41 attempted its takeoff under crosswind conditions with a runway contaminated with packed snow and patchy ice. At the approximate time of the takeoff attempt, there were crosswinds of 10-12 knots. Gusts of up to 22 knots were reported in the general area near the time of the accident. Asymmetric thrust (for example, inadequate thrust from the No. 1 engine) could have resulted in the loss of directional control experienced by flight 41. In the absence of a cross-check of other engine instruments, a malfunctioning EPR indicator could have led the flightcrew to unknowingly set inadequate thrust for the No. 1 engine. However, given the flight engineer’s recollections of evenly matched engine acceleration and consistent EPR and N1 indications from all four engines, and the absence from the CVR of flightcrew discussions of abnormal throttle alignment, the Safety Board concludes that asymmetric thrust was not a factor in the loss of directional control. Having verified the realism of the Boeing engineering flight simulator in reproducing the ground handling characteristics of the B-747 on slippery runways, the Safety Board applied the findings of its August 8, 1996, flight simulation study to the circumstances and events in this accident. In all simulations in which the pilot did not use the nosewheel steering tiller for directional control (including those conducted with icy runway conditions and winds gusting up to 40 knots), the simulated airplane was controllable along the runway centerline. In contrast, when pilots attempted to maintain the runway centerline using the tiller under slippery runway conditions with a 12-knot crosswind, a slight overcontrol at the very beginning of the takeoff roll repeatedly led to the loss of traction and steering capability from the nosewheel, followed by the loss of directional control. Given that it is very unlikely that the captain did not try to control the airplane’s tendency to weathervane into the crosswind, and given the consistent controllability of the airplane under accident conditions when the tiller was not used (during the simulation study), the Safety Board concludes that the captain’s failure to correct the airplane’s deviation from the centerline resulted from his overcontrolling the nosewheel steering through the tiller.

[PROBABLE CAUSE] Inadequate Boeing 747 slippery runway operating procedures developed by Tower Air, Inc., and the Boeing Commercial Airplane Group and the inadequate fidelity of B-747 flight training simulators for slippery runway operations contributed to the cause of this accident. The captain’s reapplication of forward thrust before the airplane departed the left side of the runway contributed to the severity of the runway excursion and damage to the airplane. 58 4. RECOMMENDATIONS As a result of the investigation of this accident, the National Transportation Safety Board makes the following recommendations: --to the Federal Aviation Administration: Require modification of applicable operating procedures published by the Boeing Commercial Airplane Group and air carrier operators of the B-747 to further caution flightcrews against use of the tiller during slippery runway operations, including low-speed operations (for airplanes equipped with rudder pedal steering) and to provide appropriate limitations on tiller use during these operations (for airplanes not equipped with rudder pedal steering). (A-96-150) Issue a flight standards information bulletin to principal operations inspectors assigned to air carriers operating the B-747, informing them of the circumstances of this accident and requesting a review and modification, as required, of each air carrier’s takeoff procedure regarding pilot hand position with respect to the tiller. (A-96-151) Require the Boeing Commercial Airplane Group to develop operationally useful criteria for making a rapid and accurate decision to reject a takeoff under slippery runway conditions; then require that B-747 aircraft flight manuals, operating manuals, and training manuals be revised accordingly. (A-96-152) Evaluate Boeing 747 simulator ground handling models and obtain additional ground handling data, as required, to ensure that B-747 flight simulators used for air carrier flightcrew training accurately simulate the slippery runway handling characteristics of the airplane. (A-96-153) After completing this evaluation, issue a flight standards information bulletin urging principal operations inspectors assigned to air carrier operators of the Boeing 747 to enhance simulator training for slippery runway operations, including limitations on tiller use and instructions for rudder use during the takeoff roll. (A-96-154) Develop certification standards for the installation of secondary galley latches; then use those standards to conduct an engineering review of secondary galley latches on all transport-category 59 aircraft.

[ANALYSIS] The Safety Board concludes that this procedural change by Tower Air will make overcontrol of the tiller less likely for its own operations; however, other air carrier operators of the B-747 may need to make similar changes to their procedures. Consequently, the Safety Board believes that the FAA should issue a flight standards information bulletin (FSIB) to POIs assigned to air carriers operating the B-747, informing them of the circumstances of this accident and requesting a review and modification, as required, of each air carrier’s takeoff procedure regarding pilot hand position with respect to the tiller. The Safety Board recognizes that it may be a natural reaction for a pilot to persevere in a takeoff attempt when faced with an apparently minor hesitation of an airplane to respond to rudder input. However, the circumstances of this accident indicate that during takeoff in a B-747 on a slippery runway, the pilot must abort at the very first indication of a directional control loss. The Boeing B-747 Operations Manual and Tower Air B-747 Flight Manual direct pilots who are performing takeoffs on slippery runways to immediately reject the takeoff if deviations from the runway centerline cannot be controlled. While this accident demonstrates the soundness of this advice, the accident also indicates that the provisions in these manuals are not adequately specific, particularly in their references to deviations that “cannot be controlled.” Tower Air’s chief of flight standards suggested a criterion for rejecting takeoffs under slippery runway/crosswind conditions that may be useful for pilot decisionmaking in the future. He linked the takeoff rejection decision to the recommended procedure of limiting rudder pedal steering input to one-half full travel to get optimal cornering friction. He indicated it was clear that if a pilot could not control the airplane with one-half rudder pedal travel, the takeoff should be rejected. This advice may be operationally useful for all B-747 pilots, if it can be verified by the FAA and aircraft manufacturer. The Safety Board concludes that current B-747 flight manual guidance is inadequate about when a pilot should reject a takeoff following some indication of a lack of directional control response. Consequently, the Safety Board believes that 45 the FAA should require Boeing to develop operationally useful criteria for making a rapid and accurate decision to reject a takeoff under slippery runway conditions; then require that B-747 aircraft flight manuals, operating manuals, and training manuals be revised accordingly. 2.2.5 Training Simulators for B-747 Slippery Runway Operations The air carrier and FAA pilots who participated in the August 8, 1996, simulation study believed that the Boeing engineering simulator had more realistic ground handling performance than the simulators Tower had provided for pilot training.

[ANALYSIS] The Safety Board’s aircraft performance study of the tire marks on runway 4L from the accident airplane (see section 1.18.7) indicated that it departed the left edge of the runway with a shallowing leftward veer. This evidence implies that the captain was beginning to regain control of the airplane when it left the runway. The simulation study results indicated that tiller inputs alone would have been incapable of this recovery of control. When interviewed after the accident, the captain recalled that he had applied increasing amounts of right rudder as the airplane veered to the left. However, based on the consistent effectiveness of rudder inputs in the simulation study and the tire mark evidence that directional control was being regained at the runway’s edge, the Safety Board concludes that the captain of flight 41 first relied on right tiller inputs as the airplane continued to veer left, then applied insufficient or untimely right rudder inputs to effect a recovery. 2.2.3 Timeliness of the Rejected Takeoff In his postaccident interview with the Safety Board, the captain stated that after noting the airplane’s failure to respond to his initial input of right rudder, and before deciding to reject the takeoff, he applied additional right rudder and tiller steering inputs. He then described 43 his attempts to reject the takeoff by retarding power to idle and applying maximum braking, right rudder, and nosewheel steering input. Thus, instead of rejecting the takeoff immediately after experiencing difficulty obtaining directional control, the captain continued to attempt to regain directional control with progressively greater rudder and nosewheel steering inputs. Because the FDR was not working, the Safety Board did not have sufficient information to measure the delay between the first indication of loss of control and the captain’s subsequent reduction of engine power. However, some measure of the extent of the delay can be gained from the simulation and performance studies. The simulation study showed that loss of directional control began at the relatively slow airspeeds when the aerodynamic rudder had not yet become effective (less than 50 knots), while the aircraft performance study showed that the accident airplane departed the left side of the runway at a relatively high speed (approximately 97 knots). The captain stated that he reduced power while the airplane was still on the runway, and that he had no recollection of subsequently reapplying power. However, the Safety Board’s CVR spectrum analysis clearly indicated that the thrust was partially reduced and then reapplied in significant amounts as the airplane left the runway. Physical evidence from the engines and flightcrew statements confirmed that the engine rpm increase recorded on the CVR was not an engagement of reverse thrust. Because the CVR ceased recording shortly after the reapplication of power to the engines, the Safety Board was unable to determine the amount of time that the airplane traveled off the runway under significant power. However, based on the spectrum analysis of engine sounds on the CVR, the Safety Board determined that the captain abandoned his attempt to reject the takeoff, at least temporarily, by restoring forward thrust.

[ANALYSIS] Given that it is very unlikely that the captain did not try to control the airplane’s tendency to weathervane into the crosswind, and given the consistent controllability of the airplane under accident conditions when the tiller was not used (during the simulation study), the Safety Board concludes that the captain’s failure to correct the airplane’s deviation from the centerline resulted from his overcontrolling the nosewheel steering through the tiller. This conclusion is supported by the captain’s statement that he added increasing amounts of tiller steering input during the loss of control sequence and departed the runway still holding full right tiller. The Safety Board was unable to determine with certainty the event that precipitated the captain’s overcontrol with tiller inputs. Simulation study results suggest that the B-747 has a tendency to react to crosswinds at very slow airspeeds with an initial, slight downwind drift. It would have been natural for the captain to have reacted to this slight 42 deviation with a tiller input, because the deviation would have occurred at a slow airspeed as the airplane was just beginning its takeoff roll. However, there could have been a number of additional reasons for why the captain applied steering inputs through the rudder or tiller at the start of the takeoff roll. These include a line-up that was slightly off the runway centerline, a wind gust, or a slight thrust imbalance from one or more engines as they accelerated to takeoff power. Still, regardless of the reason for beginning the control inputs, the simulation study indicated that the runway deviation was unlikely to have precipitated a loss of control without excessive steering inputs through the tiller. It is logical that overcontrol of the tiller on any aircraft would be more likely on a slippery runway than a dry runway, because airplane heading is less responsive to tiller inputs in slippery conditions. When a pilot makes a tiller input and does not obtain the expected reaction from the airplane, it is possible that the pilot will, at least initially, provide additional input to obtain the expected reaction. The lag in airplane response followed by additional control input could result in overcontrol of the tiller to the extent that the nosewheel exceeds its critical angle and loses traction. The simulation study also demonstrated that at least enough rudder effectiveness was obtained by 50-80 knots airspeed to shallow the simulator’s leftward veer before it departed the runway. In most simulations, directional control could be regained by timely use of the rudder. Given the effectiveness of rudder inputs in controlling heading deviations in the simulation study, the Safety Board sought to understand why the captain of the accident airplane was unable to recover directional control before the airplane departed the left side of the runway. The Safety Board’s aircraft performance study of the tire marks on runway 4L from the accident airplane (see section 1.18.7) indicated that it departed the left edge of the runway with a shallowing leftward veer. This evidence implies that the captain was beginning to regain control of the airplane when it left the runway. The simulation study results indicated that tiller inputs alone would have been incapable of this recovery of control.

[ANALYSIS] Because the CVR ceased recording shortly after the reapplication of power to the engines, the Safety Board was unable to determine the amount of time that the airplane traveled off the runway under significant power. However, based on the spectrum analysis of engine sounds on the CVR, the Safety Board determined that the captain abandoned his attempt to reject the takeoff, at least temporarily, by restoring forward thrust. The Board’s aircraft performance study indicated that as a result of the reapplication of thrust, the airplane continued to accelerate as it approached the edge of the runway. 2.2.4 B-747 Slippery Runway Operating Procedures Because the Safety Board recognized that on a slippery runway, directional control of the B-747 could be lost rapidly by overcontrol of the tiller, it evaluated the existing procedures established by Tower Air and Boeing for operating the B-747 on slippery runways. As a result of the Tower Air procedure to guard the tiller during takeoff until 80 knots, the captain was ready to use the tiller during the beginning of the takeoff roll. Tower Air and Boeing procedures urge pilots to use the rudder and rudder pedal steering during takeoff. However, B-747 procedural information produced by both the airline and the manufacturer permit the tiller to be used at the beginning of the takeoff. In its 1994 Standards Memo, Tower Air stated, “Use of the tiller is not recommended unless rudder pedal steering is not sufficient during the early takeoff roll.” Boeing stated in its Flight Crew Training Manual for the B-747, “Do not use nosewheel tiller during takeoff roll unless required initially due to crosswind.” The Safety Board is concerned that these procedures encourage use of the 44 tiller at the beginning of the takeoff roll, during which the Safety Board’s simulation study found the B-747 to be most susceptible to loss of control on slippery runways. The Safety Board concludes that current B-747 operating procedures provide inadequate guidance to flightcrews regarding the potential for loss of directional control at low speeds on slippery runways with the use of the tiller. The Safety Board believes that the FAA should require modification of applicable operating procedures published by Boeing and air carrier operators of the B-747 to further caution flightcrews against use of the tiller during slippery runway operations, including low-speed operations (for airplanes equipped with rudder pedal steering) and to provide appropriate limitations on tiller use during these operations (for airplanes not equipped with rudder pedal steering). The Safety Board was informed by Tower Air after the accident that it had reevaluated and eliminated its standard procedure of guarding the tiller during the takeoff roll through 80 knots. The Safety Board concludes that this procedural change by Tower Air will make overcontrol of the tiller less likely for its own operations; however, other air carrier operators of the B-747 may need to make similar changes to their procedures.

[ANALYSIS] The Board is concerned that air carrier B-747 pilots currently are not able to obtain needed training on slippery runway procedures, including proper tiller and rudder techniques, because training simulators have not incorporated the latest ground handling model (such as that implemented on the Boeing engineering simulator). Further, although existing flight test data on slippery runway handling characteristics are limited, the increasing use of high capacity FDRs and quick access maintenance recorders enables data on slippery runway handling to be obtained from actual line flying experience. Many B-747-400 models are equipped with these recorders. The Safety Board concludes that improvements in the slippery runway handling fidelity of flight simulators used for B-747 pilot training are both needed and feasible. Consequently, the Safety Board believes that the FAA should evaluate B-747 simulator ground handling models and obtain additional ground handling data, as required, to ensure that B-747 flight simulators used for air carrier flightcrew training accurately simulate the slippery runway handling characteristics of the airplane. The Safety Board also believes that after completing this evaluation, the FAA should issue an FSIB urging POIs assigned to air carrier operators of the B747 to enhance simulator training for slippery runway operations, including limitations on tiller use and instructions for rudder use during the takeoff roll. 2.2.6 Summary of Flightcrew Actions and Decisions The captain’s use of the tiller control for nosewheel steering during the takeoff roll, combined with his untimely or inadequate use of rudder inputs, allowed the loss of directional control to develop. As this occurred, the airplane’s deviation from the centerline and its unresponsiveness to steering inputs provided cues that, regardless of the adequacy of existing procedures and training methods, should have prompted the captain to reject the takeoff more quickly than he did. Therefore, the Safety Board concludes that the captain’s failure to reject the takeoff in a timely manner was causal to the accident. Still, better procedures for operating the B-747 under slippery runway conditions and improved ground handling fidelity of the flight simulators used for B-747 pilot training could have better prepared the captain for handling the situation that confronted the accident flight.

[ANALYSIS] Although the airport personnel claimed that the report was made, there was no documentation of a timely report in their records; the only such record was of a postaccident entry in the operations office computer. The control tower was required by FAA Order 7110.65J to advise pilots of runway friction readings when they were received from airport management, but the control tower personnel claimed that they did not receive these reports. The Safety Board 27 Federal Aviation Administration. FAA 90 Day Safety Review. Washington, DC. September 16, 1996 (mimeo). 54 was unable to determine whether the runway friction measurement data were sent or received. However, the Safety Board concludes that the failure of the PNY&NJ or FAA air traffic control tower personnel to provide these data to the pilots of flight 41 did not contribute to this accident. Although the guidance currently provided by the FAA on runway friction measurement and reporting may be helpful to airport operators, it is incomplete because friction coefficient measurements of various types are not correlated with braking performance of different airplane types or configurations. The International Civil Aviation Organization (ICAO) Guidance Material Supplementary to Annex 14, Volume I, 6, includes a table of friction coefficient measurements correlated with descriptive values, i.e., good, medium, poor. However, this table is provided for informational use only, and it, too, does not establish clearly defined parameters applicable to airplane types. The Safety Board is concerned about the frequent occurrence of veeroffs, overruns, and other related events by large airplanes when runways are contaminated with ice, snow, and/or slush (including this accident). The continuing problem with safety during ground operations is related to several problems. There is a clear need to measure the slipperiness of runway and taxiway surfaces. However, those values must then be quantified into meaningful information that pilots can use to evaluate the expected performance of their specific airplane. This would require airport operators to maintain their equipment within specific tolerances, and it would require the technicians operating the equipment to adhere to appropriate standards in using the equipment. If the FAA had been responsive to the Safety Board’s 1982 safety recommendations on this subject (see section 1.10.3), the industry might have already resolved these problems. The FAA has made considerable progress in providing and implementing procedures for airport operators to perform friction measurements during periods of ice/snow and slush contamination. However, such measurements are still not required, and there is no standardization of the equipment currently being used. Further, there are no means to compare measurement standards or translate the data into aircraft performance. A key issue is that no significant progress has been made in correlating stopping distance data from airplane manufacturers’ flight tests and calculations with the friction values obtained from measuring devices. An outcome of these correlations could be the establishment of objective standards for air carrier operations on slippery runways, perhaps extending to the establishment of appropriate minimum runway friction levels for operational use.

[ANALYSIS] Ground scar evidence from the accident site indicated that the airplane was traveling east-northeast when it impacted terrain. Thus, it was on a heading that pointed away from any relevant destination or previous flight track. An impact study that evaluated the airplane’s crush damage and the ground scar evidence determined that the airplane impacted terrain at a pitch attitude of about 17° airplane nose up and bank angle of about 30° left wing down. Also, an airplane performance study that produced a number of possible flightpath scenarios determined that flight control inputs and airplane maneuvering immediately before terrain impact were probable. The NTSB concludes that that the airplane was in a climbing left turn when it collided with terrain and that flight control inputs occurred shortly before terrain impact. Because only limited information was available, the sequence of events after the airplane’s last recorded position up to the moment of likely control input and terrain impact is unknown. FAA Advisory Circular 61-134, “General Aviation Controlled Flight into Terrain Awareness,” indicates that controlled flight into terrain (CFIT) occurs when “an airworthy aircraft is flown, under the control of a qualified pilot, into terrain (water or obstacles) with inadequate awareness on the part of the pilot of the impending collision.” Based on this definition, the circumstances of this accident could be consistent with a CFIT event if the accident pilot actively controlled the airplane but navigated inappropriately at some point after 47 NTSB Aircraft Accident Report 48 the airplane’s last known position. However, the airplane performance study indicates that the airplane also could have reached the accident site if the pilot provided little or no input on the flight controls (that is, did not actively control the airplane, which would be inconsistent with a CFIT event) for a period of nearly 3 minutes after the airplane’s last known position and then made flight control inputs only in the moments before impact. Because none of the flightpath scenarios evaluated by the airplane performance study can be considered more probable or more compelling than another, the following human performance analysis discusses factors that could affect the pilot’s possible deliberate actions or explain his possible lack of action during the final 3 minutes of the flight up to the final 4 to 6 seconds before impact. 2.3 Human Performance The accident pilot’s reported experience included decades of flying in Alaska, with several years of flying in the vicinity of Lake Nerka and recent flight operations in the vicinity of the accident area. Several pilots who were professional and personal acquaintances of the accident pilot described him as a pilot who had excellent stick-and-rudder skills, exceptional knowledge of the Alaska area and terrain, and an innate sense of situational awareness. 2.3.1 Challenges Inherent in the Alaska Flying Environment The NTSB has had a longstanding interest in the safety of aviation operations in Alaska. In a 1995 safety study,82 the NTSB noted that flight operations83 in Alaska are subject to a challenging aviation environment. Rough terrain and adverse weather can increase the risks to safe flight operations, particularly for flights conducted under VFR.

[ANALYSIS] With or without the GNS 530 GPS unit’s TAWS terrain page in use, the availability of the GPS navigational aid and the pilot’s familiarity with the area and the flight route should have provided him sufficient references to plan for avoiding terrain if he encountered IMC or another problem. If the pilot were performing at his reported typical level of proficiency and needed, for some reason, to deviate from the airplane’s usual course to the fishing camp, he likely would have been aware that a turn to the right would overfly the low tundra (allowing for maneuvering or a safe descent, as needed), whereas a turn to the left would fly directly toward high terrain. The NTSB concludes that a deliberate execution of a left turn toward the rising terrain by the accident pilot in any weather condition would require a lack of situational awareness that is inconsistent with the pilot’s reported level of proficiency, typical safety practices, and familiarity with the area. As previously discussed, from the airplane’s last known position, one possible flightpath scenario that could place the airplane at the accident site involved only a slight amount of left flight control input initiated near the airplane’s last known position. Such a scenario could result if the pilot had become spatially disoriented and allowed an uncorrected left bank to develop. For example, the “leans,” one of the most common illusions that pilots experience, can occur in degraded visibility when a pilot inadvertently rolls the airplane into a slight bank without awareness of the bank and believes that the airplane is flying straight and level.86 However, the “leans” can occur only while a pilot neglects to scan or disbelieves the cockpit instruments that provide correct orientation guidance. Such inattention to flight instruments would be unlikely for a pilot of the accident pilot’s experience level. Regardless of the pilot’s actions or intentions, the heading of the airplane at impact suggests that the pilot was largely unaware that the airplane was approaching the rising terrain. This apparent lack of awareness suggests that the accident pilot’s perceptions and/or responses became substantially impaired at some point after the airplane’s last known Sky Connect position. Using the limited information available, the NTSB explored what types of impairment could have led the pilot to allow the airplane to turn unnoticed toward the higher terrain before the final 4 to 6 seconds of the flight when he applied aggressive flight control inputs. The pilot 86 B. Cheung, “Nonvisual Illusions in Flight,” in F.H. Previc and W.R. Ercoline, eds., Spatial Disorientation in Aviation (Reston, Virginia: American Institute of Aeronautics and Astronautics, 2004), pp. 260-263. NTSB Aircraft Accident Report 52 did not have any injuries noted on autopsy that would indicate that he was on the flight controls at the time of impact, but, given the nature of the forces involved, no such injuries would necessarily be expected.

[CONCLUSIONS > FINDINGS] 3.1 Findings 1. The investigation determined that the pilot was certificated and qualified in accordance with Federal regulations. 2. Examinations of the recovered engine, propeller, and airframe components revealed no evidence of any preimpact failures. 3. The weather conditions forecasted for and observed in the area on the day of the accident did not appear to be exceptional compared to the conditions that the pilot experienced on previous flights. 4. The airplane was in a climbing left turn when it collided with terrain, and flight control inputs occurred shortly before terrain impact. 5. The airplane’s radar altimeter system provided both aural and visual altitude alerts about 4 to 6 seconds before impact, which likely prompted the pilot to take aggressive action on the flight controls, resulting in the airplane nose-up pitch and left-bank angles evident at the accident site. 6. Had the pilot not inhibited the terrain awareness and warning system’s aural voice and pop-up text alerts, the system would have provided an aural and visual alert up to 30 seconds before the impending collision. 7. A deliberate execution of a left turn toward the rising terrain by the accident pilot in any weather condition would require a lack of situational awareness that is inconsistent with the pilot’s reported level of proficiency, typical safety practices, and familiarity with the area. 8. A medical condition leading to transient incapacitation or impairment could explain the circumstances of this accident however, it is not possible to determine whether such a scenario occurred. 9. Although the pilot had some precursors for the development of fatigue, and the accident circumstances are consistent with fatigue impairment or a sleep event, there is insufficient evidence to determine whether fatigue-related performance or alertness impairments played a role in the accident. 10. The accident pilot’s recent major life events placed him at an elevated risk for stress at the time of the accident, but it is not possible to determine how, or to what extent, this stress may have affected his performance. 11. A crash-resistant flight recorder system that captures cockpit audio, images, and parametric data would have substantially aided investigators in determining the circumstances that led to this accident. 68 NTSB Aircraft Accident Report 12. Had the emergency locator transmitter remained attached to its mounting tray, it would not have become separated from its antenna, and its signals likely would have been detected soon after impact; as a result, rescue personnel would have received timely notification of the accident and its location and could have reached the survivors hours earlier, when the weather and daylight were more conducive for their evacuation. 13.

[ANALYSIS] The Sky Connect system provided five position reports for the airplane at 3-minute intervals during the accident flight. These data, when compared with the Sky Connect data available for three of the pilot’s four previous passenger-carrying flights from the lodge to the 46 NTSB Aircraft Accident Report fishing camp, showed that the accident flight’s route appeared to be the most northern and eastern route taken by the pilot by at least 1 nm (because of the interval of the data samples, the precise flight routes are not known). According to the passenger seated in the cockpit, the pilot had stated he was taking off in a different direction than usual because of weather. The wind was generally from the south-southwest, which would likely create turbulence over Lake Nerka, which was immediately downwind of higher terrain on the southwest shore of the lake. The GCI pilot with whom the accident pilot flew from the lodge to DLG that morning reported that the turbulence they encountered may have been a comfort issue for passengers, and the accident pilot, during his return trip from DLG to the lodge, reported (in the form of a PIREP) encountering “extremely irritating” turbulence. Thus, the apparent more northern and eastern positioning of the accident flight’s initial path may have been the result of an intentional pilot action to minimize the turbulence for the passengers. The Sky Connect position reports for the accident flight showed that, after taking off, the airplane was at an altitude of about 1,080 feet msl at one point and that, as it proceeded southeast, its next recorded altitudes were 780 feet msl and 920 feet msl (at the last point recorded). Flight at such changing altitudes would be consistent with the pilot maneuvering to remain in VMC below the clouds and suggests the possibility of variable cloud ceilings along the flight route. Although the initial accident flight route appeared to be offset to the north and east (when compared to the accident pilot’s previous routes to the fishing camp), the last recorded position and ground track for the airplane were such that, if that ground track were maintained, the airplane would have traveled through the wide pass west of the Muklung Hills (near the “MUKLG” GPS waypoint), which would be consistent with the pilot’s previous routes. However, the airplane instead tracked to the east and collided with terrain about 4 nm southeast of its last recorded position. Ground scar evidence from the accident site indicated that the airplane was traveling east-northeast when it impacted terrain. Thus, it was on a heading that pointed away from any relevant destination or previous flight track. An impact study that evaluated the airplane’s crush damage and the ground scar evidence determined that the airplane impacted terrain at a pitch attitude of about 17° airplane nose up and bank angle of about 30° left wing down.

[ANALYSIS] In addition, a review of the reported weather conditions at DLG during some of the pilot’s previous passenger-carrying flights from the lodge to the fishing camp revealed that the ceiling and visibility conditions reported at the time of the accident were better than what was reported at DLG during the flight performed the day before the accident. The NTSB concludes that the weather conditions forecasted for and observed in the area on the day of the accident did not appear to be exceptional compared to the conditions that the pilot experienced on previous flights. Thus, the accident pilot’s decision to depart on the accident flight appears to have been based on his own assessment that the weather had adequately improved to the extent that he felt comfortable making the flight. 2.3.2 Situational Awareness and Spatial Disorientation Because of the likelihood of localized variations, the weather conditions encountered during the accident flight are not known. Also, the passengers’ descriptions of the weather vary. Three passengers stated that the flight was conducted below the cloud ceiling in VMC; however, two of those passengers fell asleep before the time of impact. Another passenger who first stated that he had no indication of weather subsequently changed his statement to indicate that he could see only “white-out” conditions outside the airplane. Based on interviews with other pilots who had flown with the accident pilot, continuing VFR flight into instrument meteorological conditions (IMC), either deliberately or inadvertently, would be considered uncharacteristic of the accident pilot. Had the visibility and ceilings for the accident flight been 5 miles or greater (similar to that which a Cessna 208 pilot reported in the area about 30 minutes before the accident), the rising terrain at Muklung Hills would have been visible to the pilot from the airplane’s last recorded Sky Connect position. However, had the visibility been 3 miles (similar to the 1455 DLG observation of 3 miles visibility with light rain), Muklung Hills and the peaks that define the wide and narrow passes to the west would not have been visible to the pilot from the airplane’s last recorded position. Also, if a 1,000-foot overcast ceiling existed (similar to the 1455 DLG observation), the peaks of Muklung Hills would have been obscured but the floor of the tundra may have been visible. In that case, the Muklung River, which crosses the tundra and flows through the wide pass, likely would have served as a prominent landmark. Although one passenger had noted that the pilot’s visual navigational methods appeared to include following streams, it is not known whether the pilot typically used the Muklung River as a visual reference to navigate to the wide pass. 49 NTSB Aircraft Accident Report 50 The accident airplane was equipped with a variety of avionics designed to assist the pilot with navigation, situational awareness, and terrain avoidance.

[ANALYSIS] According to the airplane impact study that considered the topography near the accident site and the radar altimeter system’s selected 275-foot agl DH, the estimated elapsed time from the activation of the radar altimeter’s annunciator light and aural tone to the time of impact was about 4 to 6 seconds. The NTSB concludes that the airplane’s radar altimeter system provided both aural and visual altitude alerts about 4 to 6 seconds before impact, which likely prompted the pilot to take aggressive action on the flight controls, resulting in the airplane nose-up pitch and left-bank angles evident at the accident site. The NTSB further concludes that, had the pilot not inhibited the TAWS’s aural voice and pop-up text alerts, the system would have provided an aural and visual alert up to 30 seconds before the impending collision. Examination of the GNS 530 GPS unit revealed that, when tested, the unit’s map display was consistent with the area of the accident site, indicating that the unit was on at the time of the accident. The GPS unit’s TAWS terrain page was capable of depicting the proximity of potentially hazardous terrain on its display (inhibiting the Mid-Continent unit’s visual alerts has 84 NTSB/SS-95/03. 85 The study also noted that other VFR flight into IMC accidents can develop into a loss of aircraft control and that a GPS is unlikely to prevent such accidents. NTSB Aircraft Accident Report 51 no effect on the GPS unit’s TAWS terrain page depictions). However, the GNS 530 GPS unit’s limited memory does not store user page display settings; therefore, it is not known if the accident pilot may have used the GPS unit’s TAWS terrain page or navigation map during the accident flight. A simulation using the airplane’s GNS 530 GPS showed that, had the pilot used the GPS unit’s TAWS terrain page, nearly all of the terrain near the accident site would have been depicted in either red or yellow. Testing of the GPS unit showed that, had the pilot used only the navigation map page (not the terrain page), the navigation map would have displayed “MUKLG” on the map, indicating the location near the Muklung River (visual landmark) within the wide pass near Muklung Hills relative to the airplane’s position. Examination of all the user-defined waypoints programmed into the GPS found none that, if navigated toward from the airplane’s last known position, would have led the pilot to the accident site vicinity. With or without the GNS 530 GPS unit’s TAWS terrain page in use, the availability of the GPS navigational aid and the pilot’s familiarity with the area and the flight route should have provided him sufficient references to plan for avoiding terrain if he encountered IMC or another problem.

[ANALYSIS] In a 1995 safety study,82 the NTSB noted that flight operations83 in Alaska are subject to a challenging aviation environment. Rough terrain and adverse weather can increase the risks to safe flight operations, particularly for flights conducted under VFR. In the study, the NTSB noted that weather in Alaska can be quite variable depending on the climatic zone and time of year and that innumerable localized climatic conditions exist near mountainous terrain, mountain passes, and glaciers. The accident pilot did not receive an FAA weather briefing before the accident flight. Evidence from the lodge computers indicates that someone (possibly the accident pilot, given the aviation weather-oriented websites visited) checked online weather information between 0740 and 0745 (before the pilot’s morning flight to DLG) and again between 1132 and 1133 (after the pilot returned from DLG). Accessed weather information that may have been relevant to the accident flight included a PADL TAF that indicated that conditions for DLG were forecasted to include (from 1100) wind from 160° at 11 kts, visibility 5 miles with light rain showers and mist, and ceiling overcast at 900 feet. During the accident pilot’s morning flight from DLG, he reported (in a PIREP) encountering ceilings at 500 feet and 2 to 3 miles visibility in light rain. After completing that flight, the pilot decided that the weather was not conducive for a trip to the fishing camp; however, the weather reportedly improved after lunch, and the pilot informed the guest party 82 See Aviation Safety in Alaska, Safety Study NTSB/SS-95/03 (Washington, DC: National Transportation Safety Board, 1995). 83 The study focused primarily on commercial flight operations (particularly air taxi operations conducted under the provisions of 14 CFR Part 135) but noted that many of the factors affecting the safety of Part 135 operations also affect general aviation (Part 91) operations. NTSB Aircraft Accident Report co-host that a trip to the fishing camp could be made if the guests wanted to go. There was no indication that GCI placed pressure on the pilot to make the flight or to complete it to the destination. Although GCI personnel and some passengers stated that they thought that the pilot checked the weather during lunch, no evidence of dining hall computer use corresponded with that timeframe. However, there was no indication that the area weather conditions on the day of the accident were particularly concerning for either the accident pilot or the GCI president, who chose to depart the lodge in his airplane (about 35 minutes after the accident flight departed) on a pleasure flight with his wife. In addition, a review of the reported weather conditions at DLG during some of the pilot’s previous passenger-carrying flights from the lodge to the fishing camp revealed that the ceiling and visibility conditions reported at the time of the accident were better than what was reported at DLG during the flight performed the day before the accident.

[ANALYSIS] Workload was further increased by the limited support he received from ATC, which included unresponsiveness, inefficient clearances to ADW, and an inability to provide a requested ASR approach. The NTSB therefore concludes that the pilot‘s workload increased substantially and unexpectedly as a result of encountering instrument weather conditions. NTSB Aircraft Accident Report 55 Further, the NTSB concludes that the pilot‘s expectation that he could descend below the cloud ceiling at an altitude above the MDA for the approach, his familiarity with ADW, and the reduction in workload a return to visual conditions would have provided are all factors that may have encouraged the pilot to deviate below the glideslope and attempt to ―duck under‖ the cloud ceiling. Regardless of the pilot‘s expectations for reaching VMC, once he deviated below the glideslope, he should have been prepared to level off at the MDA prescribed by the instrument approach procedure until he was able to confirm the airport environment as required by federal regulations. The pilot‘s failure to stop or slow the descent indicates that he was not aware of the helicopter‘s excessive descent rate or its height above the terrain, likely because he was looking outside the cockpit for the ground. At night, with low cloud ceilings, reduced visibility, and no surface light references, there were insufficient visual cues available for the pilot to establish ground reference. The helicopter was equipped with a radar altimeter, which should have alerted the pilot when he descended below 300 feet agl, about 6 seconds before impact with the trees.72 However, there was no decrease in the helicopter‘s descent rate after it passed through 300 feet agl. Testing of the radar altimeter revealed no discrepancies that would have prevented it from functioning normally during the accident flight. The NTSB concludes that the pilot failed to adhere to instrument approach procedures when he did not arrest the helicopter‘s descent at the MDA. The NTSB further concludes that although descent rate and altitude information were readily available through cockpit instruments, the pilot failed to monitor the instruments likely because he was preoccupied with looking for the ground, which he could not identify before impact due to the lack of external visual cues. Additionally, NTSB investigators asked a manufacturer of TAWS to determine what pilot alerts would be expected if the helicopter had been equipped with TAWS. The manufacturer ascertained that three aural terrain alerts would have been generated at 7, 4, and 2 seconds prior to tree impact, and an aural glideslope alert would have been generated 24 seconds prior to tree impact if a valid glideslope signal was being received. It is unlikely the glideslope warning would have caused the pilot to arrest his descent since it appears that he intentionally deviated from the glideslope. However, if the helicopter had been equipped with a TAWS, the aural terrain alerts of ―Caution Terrain,‖ ―Warning Terrain,‖ and ―Pull-up,‖ would have been provided.

[CONCLUSIONS > FINDINGS] The lack of information regarding the accident airplane‘s navigation frequency settings and flight instrument indications precluded National Transportation Safety Board investigators from determining why the pilot believed he was not receiving a valid glideslope signal. 8. The pilot‘s workload increased substantially and unexpectedly as a result of encountering instrument weather conditions. 9. The pilot‘s expectation that he could descend below the cloud ceiling at an altitude above the minimum descent altitude for the approach, his familiarity with Andrews Air Force Base, and the reduction in workload a return to visual conditions would have provided are all factors that may have encouraged the pilot to deviate below the glideslope and attempt to ―duck under‖ the cloud ceiling. NTSB Aircraft Accident Report 72 10. The pilot failed to adhere to instrument approach procedures when he did not arrest the helicopter‘s descent at the minimum descent altitude. 11. Although descent rate and altitude information were readily available through cockpit instruments, the pilot failed to monitor the instruments likely because he was preoccupied with looking for the ground, which he could not identify before impact due to the lack of external visual cues. 12. If the helicopter had been equipped with a terrain awareness and warning system, aural terrain alerts of ―Caution Terrain,‖ ―Warning Terrain,‖ and ―Pull-up,‖ would have been provided. These would have been more salient than the alert provided by the radar altimeter and likely would have caused the pilot to attempt to arrest his descent. 13. The failure of the Potomac Consolidated Terminal Radar Approach Control controller to provide the current Andrews Air Force Base weather information likely led the pilot to expect that he could descend below the cloud ceiling and establish visual contact with the ground at an altitude well above the minimum descent altitude for the approach. 14. Air traffic services provided by the Ronald Reagan Washington Airport Tower and Potomac Consolidated Terminal Radar Approach Control controllers to the accident flight exhibited numerous procedural deficiencies, including unresponsiveness, inattention, and poor radar vectoring. These deficiencies were a distraction to the pilot and increased his workload by requiring him to compensate for the poor services provided. 15. Although the pilot met the recent-experience requirements to act as pilot-in-command under instrument flight rules, he was not proficient in instrument flight. This lack of proficiency likely contributed to the pilot‘s failure to properly conduct what effectively became a nonprecision approach at night in instrument conditions. 16. Changes made by the Maryland State Police Aviation Command to its instrument training program about 10 months before the accident did not promote instrument proficiency. 17.

[ANALYSIS] He could have declared an emergency, which would have prompted the ADW controller to provide assistance, possibly including the ASR approach. Also, he could have executed a missed approach and attempted the ILS approach a second time to determine if the glideslope failure was a perceived failure or a legitimate one. Additionally, there were 11 other instrument approaches at ADW, any of which he could have requested. About 27 seconds after the controller stated that she was unable to provide an ASR approach, upon reaching an altitude of about 1,450 feet msl on the glideslope and at a distance of about 4.0 miles north of the runway threshold, the helicopter‘s rate of descent increased rapidly from about 500 feet per minute (fpm) to greater than 2,000 fpm. The helicopter continued the descent, passing through the MDA for the localizer approach (407 feet agl), the alert height set on the radar altimeter (300 feet agl), and the decision height for the ILS approach (200 feet agl), before impacting trees and terrain about 3.2 miles north of the runway threshold. Data recovered from the PAR computer indicate that the helicopter impacted with the engines near idle power, the main rotor system at 100 percent rpm, and an indicated airspeed of about 92 knots. No evidence was found to indicate that the pilot made any attempt to arrest the helicopter‘s descent before impact. The sudden increase in the helicopter‘s rate of descent may indicate that the pilot deliberately deviated from the approach procedure by attempting to ―duck under‖ the cloud layer to regain visual conditions. Several factors might have encouraged the pilot to attempt to ―duck under‖ the cloud ceiling. First, the outdated ADW weather report provided by the PCT controller indicated a ceiling of 1,800 feet and 7 miles visibility. Given these conditions, the pilot would have had a reasonable expectation that he could have descended below the cloud ceiling at an altitude well above the MDA for the approach. Second, the pilot was based at ADW and familiar with its visual environment. This likely would have given him confidence that he would be able to identify landmarks and quickly determine his position once he descended below the cloud ceiling. Third, a return to visual conditions would have relieved the pilot of the additional workload generated by instrument flight. All medical flights involve significant workload and time pressure to provide patients with timely service to medical aid. However, this flight involved unusually high workload once the pilot encountered instrument conditions. These conditions greatly increased the demands on the pilot since he had little recent experience in actual instrument flying. Additionally, the weather prevented him from reaching the planned medical center, forcing him to divert with the new responsibility of considering alternate transport arrangements for the patients along with his diversion considerations. Workload was further increased by the limited support he received from ATC, which included unresponsiveness, inefficient clearances to ADW, and an inability to provide a requested ASR approach. The NTSB therefore concludes that the pilot‘s workload increased substantially and unexpectedly as a result of encountering instrument weather conditions.

[ANALYSIS] Review of ADW weather data showed that the weather given to the pilot had been issued at 1855 local time, almost 5 hours earlier. Weather information available to controllers is time-stamped, and the controller should have noticed that the report was outdated. The controller did not appear to have noticed the time discrepancy and did not advise the pilot of the age of the report or take any action to contact ADW for current weather information. During the time that the aircraft was being vectored for the approach, ADW was reporting wind from 110° at 3 knots, visibility 7 miles, cloud ceiling 1,300 broken, temperature 20° C, dew point 20 °C. Although these weather conditions were well above the localizer-only and ILS landing minimums for the runway 19R approaches, the cloud ceiling was 500 feet lower than that reported to Trooper 2. If the pilot had been given the current weather information, he would have expected to be unable to see the ground until he had reached a significantly lower altitude. Knowing that the ceiling was 500 feet lower may have discouraged the pilot from attempting to duck under the cloud ceiling and resulted in his continuing the approach to the airport at a 500 fpm rate of descent. Therefore, the NTSB concludes that the failure of the PCT controller to provide the current ADW weather information likely led the pilot to expect that he could descend below the cloud ceiling and establish visual contact with the ground at an altitude well above the MDA for the approach. The controller did not explicitly issue Trooper 2 an instrument clearance or an IFR transponder code. An IFR transponder code is necessary for aircraft to receive MSAW service. When interviewed, he stated that the pilot never requested an instrument clearance, and the controller believed that the pilot wanted a VFR practice approach to ADW. As the pilot had requested IFR from DCA and there was no further discussion of VFR operations by either the pilot or the PCT controller after he was handed off, the pilot could have reasonably expected that he was operating under IFR despite not having received an explicit IFR clearance from PCT. The NTSB Aircraft Accident Report 57 controller's statement that he believed Trooper 2 was VFR all the way to touchdown is contravened by the controller's failure to restrict the aircraft to VFR operations as required by FAA directives. Between the time Trooper 2 contacted PCT and issuance of the IFR approach clearance, Trooper 2‘s status as an IFR or VFR flight was ambiguous. Once the helicopter was cleared for an approach without a VFR restriction, Trooper 2 was an IFR flight and should have been given an IFR transponder code.

[ANALYSIS] NTSB investigators considered two possible reasons for the pilot‘s request for a surveillance approach. One possibility was that the pilot did not have the ILS/LOC runway 19R approach chart visible as he was executing the approach to ADW. Therefore, he would not have been able to reference it to determine the MDA, the missed approach point, and other details of the localizer approach. Under this circumstance, the pilot‘s request for an ASR approach would be plausible because, if a surveillance radar approach had been provided, the controller would have advised the pilot that the MDA for the surveillance radar approach was 780 feet msl and instructed the pilot when to begin descent to the MDA. Additionally, at each mile of the final approach, the controller would have provided the pilot with the helicopter‘s distance from the missed approach point. According to MSP Aviation Command personnel, the instrument approach charts were kept in a pouch on the right side of the pilot‘s seat and would have been readily accessible to the pilot. At the accident site, the charts were found scattered throughout the debris near the nose of the helicopter, precluding a determination of whether the pilot was looking at a particular chart while on approach to ADW. Another possible reason for the pilot‘s request for an ASR approach was that he did not have the DME transceiver set to the frequency of the ADW VORTAC. The localizer approach to runway 19R requires DME readings off the ADW VORTAC to determine the final approach fix (where descent to the MDA begins) and the missed approach point (where the pilot has to decide whether to continue to land or to abort the approach). As described in section 1.16.2, the DME transceiver and the navigational receiver are usually set to the same frequency. Thus, in order to switch to the localizer frequency for runway 19R and retain DME information from the ADW VORTAC, the pilot would have had to select the ―HOLD‖ function on the navigation control head before switching frequencies. If the pilot did not select ―HOLD‖ before switching frequencies, the DME transceiver would have also been switched. The pilot would not have had any DME information and, therefore, would have had no means of determining the final approach fix or the missed approach point. Under this circumstance, the pilot‘s request for an ASR would again be plausible. NTSB Aircraft Accident Report 54 There was insufficient information available to determine whether either of these two scenarios was the reason that the pilot requested an ASR approach. Regardless of the reason for his request, once the controller denied it, the pilot still had many options available to conduct a safe landing in instrument conditions. He could have declared an emergency, which would have prompted the ADW controller to provide assistance, possibly including the ASR approach. Also, he could have executed a missed approach and attempted the ILS approach a second time to determine if the glideslope failure was a perceived failure or a legitimate one. Additionally, there were 11 other instrument approaches at ADW, any of which he could have requested.

[ANALYSIS] This could have led the pilot to conclude the weather was better than reported. Regardless of the unavailability of the DoD surface weather observations, the pilot did have access to information that indicated the weather conditions were continuing to deteriorate throughout the evening. In addition to surface observations, the pilot could also have viewed terminal area forecasts and AIRMETs via the HEMS weather tool. He could have obtained the amended terminal forecast for DCA, issued at 1933, which indicated that weather conditions 71 Local (on airport) dissemination of the ADW observations was not affected by the DoD data communications failure. NTSB Aircraft Accident Report 50 were expected to deteriorate temporarily from VFR to IFR between 2000 and 2300 with the ceiling broken at 800 feet, overcast at 1,500 feet. Additionally, he could have obtained an AIRMET for IFR conditions issued at 2245 for an area that encompassed the entire route of Trooper 2. The forecast ceilings and visibilities in the DCA terminal forecast and in the AIRMET were above the MSP Aviation Command minimums for acceptance of a night medevac flight. However, the amendment to the DCA forecast and the issuance of the AIRMET indicated that weather conditions were continuing to deteriorate. Since HEMS weather tool data is not archived, it is not possible to determine what information was viewed by the pilot. It appears that the pilot based his decision to launch solely on the weather observations at College Park and DCA and the suitable conditions implied by the other medevac helicopter‘s completed flight. Other pertinent weather data—the low temperature/dew point spreads at ADW and College Park, the AIRMET for IFR conditions encompassing the route of flight, and the continuing deterioration of the weather conditions as the evening progressed—were either discounted by the pilot or not obtained. If the pilot had thoroughly obtained and reviewed all of the available weather information, it is likely he would have realized that there was a high probability of encountering weather conditions less than MSP minimums on the flight and this would have prompted him to decline the flight. Therefore, the NTSB concludes that the pilot‘s decision to accept the flight, after his inadequate assessment of the weather, contributed to the accident. According to the safety officer, at the time of the accident, MSP did not have a formal risk management program in place. He explained that there was optional guidance available to pilots in the form of a ―Risk Assessment Matrix.‖ However, review of the MSP Operations Manual revealed that it stated the flight crew ―will apply‖ the matrix and based on the risk assessment, increase visibility and ceiling minimums ―to the crew‘s comfort level prior to accepting the mission.‖ The matrix indicated that a temperature/dew point spread of less than 2° C raised the flight risk from low to medium risk. Although the matrix indicated that no flights were to be made if the risk level was high, it provided no instructions concerning medium-risk flights. There is no evidence indicating that the accident pilot consulted the matrix before the flight.

[ANALYSIS AND CONCLUSIONS] These conversations covered a number of subjects, from politics to used cars, and both crewmembers expre .ed strong views and mild aggravation concerning the subjects dise cus ed. The Safety Board believes that these conversations were distr.ctive and reflected a casual mood and lax cockpit atmosphere, which continved throughout the remainder of the approach and which contributed to the accident. The overall lack of cockpit discipline was mantfested in a number of respects, as discussed below, where the flightcrew failed to adhere to recomnanded or required procedures, At 0732:13, as the flight intercepted the inbound VOR radial for the approach, the flighterew commenced a discussion of Carowinds Tower, which was located ah2ad and to the left of the projected Zlightpath. This discussion laeted 35 seconds, during which 12 remarks wer? made concerning the subject. It is apparent that, ducing this discussion, a considerable degree of the flightcrew's attertion vas directed outside the cockpit. This particular distraction assumes significance because, during this period, the aircraft descended through 1,800 feet (1,074 feet above touchdown elevation), the altitude which should have been maintained until it crossed Ross Intersection, the final approach fix (PAF), At the end of the 35-second period, the aircraft was still 1.5 nmfi short of the FAF. It is noteworthy that at 0732:41, during the latter part of the discussion regarding Carowinds Tower, the terrain warning alert sounded in the cockpit, signifying that the aircraft was 1,000 feet above tthe ground. This warning should have been particularly significant to the flightcrew, if heeded, since it would have mae them aware that the aircraft had prematurely descended through the FAF crossing altitude of 1,074 feet above touchdown elevation, Obviously, the crew was not so alerted, since the des« cent continued. Based on pilot testimony taken at the hearing, it appeara that the crew's disregard of the terrain warning signal jn this instance may be indicative of the attitudes of many other pilote who regard the signal as more of a nuisance than a warning. If this is indeed the case, the Board believes that airline pilots should reexamine their attitudes toward the terrain warning alert, lest the purpore for which the device was installed be defeated. Although the repetitious sounding of the alarm may have a tendency to undermine its effectiveress, this acci- aioe f et i EMM Soe Me es . ie eee i + I ae 5 ~ 2 ga se ag eye ve la ety a oe .* Lind eee ee.

[ANALYSIS AND CONCLUSIONS] No such callout was made, nor was the required callout made when the plane descended through an altitude 100 feet above the MDA of 394 feet above the field elevation. The descent rate, after passing Ross, in# creased to 800 feet per miru a, where it stabilized until approximately 7 to 8 seconds prior to im... ., when it steepened considerably, The Board hac -en unable to determine the precise ceason for the almost total lack o .tt::"4 awareness on the part of the crew throughout 7 Subsequent to th? - lent, Eastern amended its procedures to require that, when the te .::. warning signal sounds, the callout at 1,000 feat abova airpor. i ation will be made, Another requirement made by Eastern ie tha. © adfo altimater will be set et MDA or at 500 feat when the landing .u baing made on runways not served by an ap-= proach procedure, 13/ Weaalso note that th: acormended maneuvering speed for 15 degrees of . flaps, which had been extended saveral minutes previously, is 160 knots. ee ~ 17 « the approach, It is possible that the crew, beceuse of the extended dura~ tion of flight in VMC >ove a low, patchy fog bank through which intermittent ground contact way possible, may have relaxed their instrument scan and relied more heavily upon visval cues to fly the approach, Such a possibility is consistent, not only with the discussion of Carowinds Tower described above, but also with the captain's remark, shortly before in pact, that "All we got to do is find the airport," and the firet officor'e response of “Yean."’ Ultimately, when the aircraft penetrated the densa fog around the eccident site, visual reference would hava been lost and a switch to instrument flight would not have been possible within the availeable tima, Tha mount likely explarction of why Flight 212 was unable to establish visual contact with tha runway environrent, whereas other flights were able to do so and thereby complete the approach, is that Flight 212, flying at a lower altitude, initially entered the fog bank at a point farther from the runway threshold and thus had a greater slante range distance through shich to sight the runway markings through the fog. Another possible reason for the crew's lack of altitude awareness involves the interrelationship between QNH (above sea level) and QFE (above field elevation) altitudes during the approach. --- Footnotes: [13/ Weaalso note that th: acormended maneuvering speed for 15 degrees of]

[ANALYSIS AND CONCLUSIONS] The captain may likewiea have beileved the aircraft was 1,000 feet higher above the field elevation than it actually was, which would mean that, in his mind, the plane never reached MDA or 100 feet above MDA, which would further explain why these calloute were naver mide. Additionally, the captain may have failed to detect tia discrepancy between the p:escribed and actual aititudes be« cause of his preoccupation with the checklist and with looking outside the cockpit, It should be emphasized that the possible explanation discussed imnediately above is based not only on evidence that is tenuous, at beat, but aleo on the inferences to be drawn from such evidenc. as to what thought processes were evolving in the mlnds of the flightcrew. Obviously, such an explanation is, to a considerable degree, speculative in nature. It is nevertheless the intent of the Board that, by including this discussion in the report, pilots will be alerted against the possibility of lapsing into such a pattern when utilizing a QFE altimeter setting procedure. We also hasten to add that, even if it is assuned that the sequence of events described in the above discussion in fact occurred, this should be taken to reflect adversely not on Eastern's system, but rather on the flightcrew's implementation of that system in this instance, By vixtue of training, experience, cockpit instrumentation, navigational aids, and approach plates, this crew was well equipped to accomplish the approach to Charlotte safely, and there is no causal factor beyond the flightcrew it~ self which would account for their failure to do so. This accident exenplifies the absolute necessity of strict adherence to prescribed procedures, particularly those pertaining to altitude awareness, during an instrument approach, Survivability Three major factors made this a partially survivable accident: {{. The occupiable area of the cabir. was compromised when the fuselage broke up, 2, The intense postimpact fire consumed the occuplable area of the tail section and the entire center section of the cabin, 3, The occupant restraint systam failed in many instances, ° evon though crash forces were within human tolerances. The cockpit, area and the forward cabin were demolished by inact with tree, The tail section, which included the last f£ve rows of passenger s@its, is classed as a survivable crea, Howaver, postcrash fire created & mjor survival problem in this section. Bodies of most of the aircraft occupants were found outside two of the major; sections of cabin wreckage, which indicates that the passenger i ; é Were hareoy. tee peer een enter aE Rae or ee Rt ap mae a ne ee aa a ~ 19 « restraint system was disrupted in these sections during cebin disintegration.

[(a) FINDINGS] Fo ee a AON OC OG SI te eRe tent NPN SEM RE UES TT Me GES REE NR DRG Mea RR, MONE a Nye ge ny > « 20 ~ The weather in the Charlotte area was characterized by shallow, patchy ground fog such that VMC existed above the fog bank, but that visibility was drastically reduced within the fog. Tha approach was flown manually by the firet officer, while the captain handled radio transmissions and accomplished ehecklist items, # ca . act tit a alt tnt eee pide eines 4 een en en The extraneous conversacion conducted by the flightcrew during the descent was symptomatic of a lax atmosphere in the cockpit which continued throughout the approach. The terrain warning alert sounded at 1,000 feet above the ground but was not heeded by the flightcrew, The aircraft descended through the final approach fix altitude of 1,800 feet more than 2 miles before the final approach fix was reached at an airspeed of 186 knots. The aircraft passed over the final approach fix at an altitude of 1,350 feat (or 450 feet below the prescribed crossing altitude) and at an airspeed of 168 knots, as compared to the Veer speed of 122 knots, Required callouts were not mide at the final aoproach fix, at an altitude of 500 feet above field elevation, ov at 100 feet above the minimum des. ent altitude, A severe postimpact fire occurred immediately after the initial impact. 10, Fatal injuries were caused by impact and thermal tvauma, 11. The door exits, axcept for the auxiliary exit in the tail, were blocked externally, 12, Doubleknit polyester clothing increased the severity of burns, (b) Probable Cause The National Transportation Safaty Board determi ies that the probable cause of the accident was the flighcarew's Lack of altitude awareness at critical points during the approach due to poor cockpit discipline in that the craw did not follow prescribed procedures, - 21» 3. RECOMMENDATIONS On October 8, 1974, tha Board fasued two safety recommendations to the PAA (A-74-85 and A-74-8) to initiate ways and means to improve professional standards among pilots. hese recommendations cited five previous air carrier approach accidents as examles of a cael acceptance of the flight environment, and added that the Charlotte ciush "reflects once again serious lapses in expected professional conduct." The PAA agrees with both recommendations and is in the process of establishing a working lfaison on this subjact with beth airline management ad air carrier pilot organizations, FS 4 SPR Scams: Ney Areata EOE? ne ORI TE HRS NEE rid ED Weiter I NOT BY THE NATIONAL TRANSPORTATION SAFETY BOARD /s/ JOWN H, REED Chairman /s/ FRANCIS H.

[ANALYSIS AND CONCLUSIONS] Although the repetitious sounding of the alarm may have a tendency to undermine its effectiveress, this acci- aioe f et i EMM Soe Me es . ie eee i + I ae 5 ~ 2 ga se ag eye ve la ety a oe .* Lind eee ee. Me <a ee eee a en eee ee Rieti na Ree eee eet oe ny » 16 = dert points up the importance of devices designed to ent ince altitude awareness at critical points in an instrument approach, 12/ Within seconds after the discussion of Carowinds Tower terminated at 0732:48, tha rate of descent of the aircraft was slowed from about 1,500 feet per minute to less than 300 feat per minute. Such a reduction in the descent rate may have bean a raflection of the switch of the first officer's attention from outeide the cockpit to the instrument panel. Prior to the reduction in the rate of descent, tha airspeed had increased to i88 knots, which clearly seems excessive in view of the f4ct that the flight had approached to within a mile of tha FAF. 13/ As tha rate of descent decreased, the airspeed also decreased, from 188 knote to 168 knots, At 0733:24, the aircraft passed over Ross {{ntersection (the FAF) at an altitude of 1,250 feet (624 feet above field elevation), which is 450 feet below the prescribed crossing altitude, The captain did not make tha required callout at the FAF, which should have included the altitude (above field elevation), deviation from the "Buy" or Vre¢ speed, and the xesult of the flag scan, Although shortly before crossing the FAF, one of the pilots stated "three ninety four," such statement obv:- ously was not a callout of the altitude, but rather a refecence to the MDA in height: above field elevation. While in the vicinity of Ross Intersection, the first officer asked for 50 degrees of flaps; this request was carried out by the captain, The airupeed at this time was 168 knots, as contrasted with the recom nendec precedura which calls for the airspeed when parsing over the FAF to be in the area of Vref, which in this instance was 122 knote. This discrepancy is a further manifestation of the overall unstabilized tature of the approach, Shortly after passing Ross Intersection, the aircraft passed through an altitude of 500 feet above field elevation, which should have prompted the captain to cali out altitude, deviation from "bug" speed, and rate of dascert. No such callout was made, nor was the required callout made when the plane descended through an altitude 100 feet above the MDA of 394 feet above the field elevation. --- Footnotes: [13/ Weaalso note that th: acormended maneuvering speed for 15 degrees of]

[ANALYSIS] In reduced visibility weather conditions, Promech pilots typically fly southwest over Ella Lake and proceed beyond the south end of the lake before turning west. This route allows the flights to avoid the high terrain located west of the lake, which includes two similarly shaped mountains. (See figure 16.) Figure 16. The typical flight route (white) passes two similarly shaped mountains before turning west. For comparison, the path flown by the accident pilot is shown in green. Although the accident pilot climbed the airplane to an altitude that would have provided safe terrain clearance along the typical route, he deviated from that route and turned the airplane west after it passed only the first of the two mountains. The pilot’s route deviation placed the airplane on a collision course with the 1,900-ft second mountain, which it struck at an elevation of about 1,600 ft msl. (See figure 17.) NTSB Aircraft Accident Report 47 Figure 17. The path flown by the accident pilot (green), viewed from the approximate altitude of the accident airplane, is shown turning to the west (right) after passing the first mountain. Note: A translucent mask is imposed on the image to provide a notional representation of reduced visibility weather conditions. The vantage point of the image distorts the flight route depiction in the foreground (where the segment of the route originates in the bottom center of the image); there is no steep vertical climb in the flight route at the location shown in the foreground. Research suggests that, when navigating, people find their way through large-scale environments by learning to identify key landmarks (route knowledge) and developing map-like internal representations (cognitive maps) of the area. When traveling, people use the landmarks to update their position on their internal cognitive map and decide when and where to turn (Golledge 2010). However, a number of factors can increase the likelihood of navigational errors. Navigational errors are more likely if landmarks are difficult to discern or if cognitive maps are inaccurate (Golledge 2010). Landmarks are more difficult to discern when visibility of fine detail is degraded or when landmarks are viewed from an unfamiliar vantage point (Cuqlock-Knopp and Whitaker 1993; Schreiber et al. 1998). Several factors were present during the accident flight that increased the likelihood that the pilot would make a navigational error.35 Visibility was degraded over the south end of Ella Lake; thus, the pilot may not have been able to clearly distinguish some of the landmarks by which he typically navigated. In addition, the accident pilot was flying the route at a relatively low altitude because of the poor weather conditions, which resulted in his viewing terrain from a lower-than-normal vantage point. (The accident pilot had transited the area earlier that day about 2,000 ft msl, and other Promech pilots had flown about 2,500 to 3,000 ft msl.) Further, the accident 35 In addition, some pilots reported that it could be difficult to see directly ahead in the DHC-3 because the turbine engine conversion resulted in less efficient clearance of rain from the windscreen. However, investigators were unable to independently evaluate these claims.

[ANALYSIS] However, investigators were unable to independently evaluate these claims. NTSB Aircraft Accident Report 48 pilot’s relatively short time flying in the local area suggests that his cognitive map of the terrain between Rudyerd Bay and Ketchikan may not have been as detailed or accurate compared to those of other pilots who had more local flying experience. Considering all these factors, it is possible that the accident pilot mistook the first mountain for the second one and turned the airplane west too early; such a mistake, in which a pilot does not recognize or maintain the desired position relative to the external ground and airspace environment is an example of geographic disorientation (Antuñano, Mohler, and Gosbee 1989). The fact that the pilot subsequently continued the flight straight and level at an altitude that would have been sufficient to clear terrain on the typical route (but was insufficient to clear the high terrain he was flying toward) further supports the possibility that the pilot was geographically disoriented. Thus, the NTSB concludes that the pilot’s continued flight in low-visibility conditions at a lower-than-normal vantage point in an area in which he lacked extensive flying experience reduced his ability to visually identify landmarks and resulted in a navigational error due to geographic disorientation. 2.2.3 CFIT According to recorded flight data, the airplane’s final turn to the west occurred about half-standard rate; then the airplane continued on a nearly straight and level flightpath toward the higher terrain for about 30 seconds. In the final 2 seconds of the flight, the airplane pitched up rapidly, experiencing a vertical acceleration of about +2 Gs before colliding with terrain. The timing of the pilot’s aggressive pitch-up maneuver strongly supports the scenario that the pilot continued the flight into near-zero visibility conditions and, as soon as he realized that the flight was on a collision course with the terrain, he pulled aggressively on the elevator flight controls in an ineffective attempt to avoid the collision. Effective CFIT avoidance requires assessment of weather conditions and appropriate aeronautical decision-making in response to cues associated with deteriorating weather. The pilot missed opportunities to prevent the accident both in his decision to take the short route despite the presence of low clouds and obscured terrain, to turn toward an area of the lake where visibility was further reduced, and to continue the flight into deteriorating conditions rather than escape them.

[ANALYSIS] However, the DO (who provided the CFIT-avoidance training) and a Promech pilot who completed the training with the accident pilot differed in their assessments of the accident pilot’s performance. Although the DO believed that the accident pilot performed well during training, the other Promech pilot said that the accident pilot experienced difficulty performing the CFIT-avoidance maneuver. Based on this conflicting information, the investigation cannot draw a conclusion about the pilot’s level of proficiency and confidence in performing the CFIT-avoidance maneuver. During tour flights, the accident pilot’s responses to actual encounters with adverse weather varied under different circumstances. Although the DO and other Promech pilots described the accident pilot as a conservative decision-maker who was willing to turn back when weather conditions were poor, the accident pilot’s flights on the day of the accident demonstrated examples of both conservative and risk-taking decisions. The accident flight occurred during the third round of Promech tours that day. ADS-B data and passenger interviews showed that, when the accident pilot was returning from his first tour on the day of the accident, he responded to an IMC encounter by executing a climbing, 180° turn and then following a different route. During the accident pilot’s second tour of the day, however, he continued his flight into deteriorating conditions, and his airplane descended as low as 375 ft agl (which was below the FAA minimum altitude for that area). In addition, during the accident flight (the pilot’s third tour of the day), he chose to fly the short route despite the presence of low clouds and obscured terrain and then continued the flight into deteriorating weather rather than escape it. One major difference between the scenarios is that, during the tour in which he reversed course, the accident pilot was flying alone, whereas, during the two tours in which he continued flight into deteriorating weather (his second tour and the accident tour), he was following 5 to 10 minutes behind other tour airplanes. This suggests that operational factors may have influenced the pilot’s decision-making. (Operational factors are discussed in section 2.3.) 2.2.3.2 Nuisance Alerts from Class B TAWS during Tour Flights The accident airplane’s Chelton FlightLogic EFIS provided the Class B TAWS capabilities required by 14 CFR 135.154(b)(2) for the airplane (based on its engine and seating configuration). Class B TAWS alerting parameters, which are specified in FAA TSO-C151c, are based on terrain clearance of 700 ft agl in cruise flight and 500 ft agl during descent. When these clearances are not NTSB Aircraft Accident Report 51 maintained, the Chelton system’s Class B TAWS provides color-coded cautions and warnings of terrain on the moving map on the MFD, a forward-looking view of terrain ahead of the aircraft (a replica of a day VFR view out the front window) on the PFD, and CWA auditory and flag alerts.

[ANALYSIS] The NWS area forecast valid for the accident area for the period before and during the accident flight included an AIRMET advisory for mountain obscuration with MVFR and temporary IFR conditions due to light rain, mist, and low ceilings. There was no record of the accident pilot obtaining any official preflight weather briefing and or reviewing any FAA weather camera images or other weather information sources before the flight. Although Promech pilots used local traffic advisory frequencies to communicate with each other (and other operators’ pilots) via radio, the airplane was not equipped with any recording device; thus, any information the accident pilot may have heard over the radio frequencies while en route is not known. The flight scheduler who was on duty the day of the accident did not recall having any preflight conversations with the accident pilot about weather. (Section 2.4.1 provides more information about the flight scheduler’s duties.) Other pilots’ descriptions of the weather they encountered, as well as digital photographs and videos taken by passengers (including those on board the accident flight), provided information about weather conditions on the short route. The Promech pilot whose flight transited Ella Lake on the short route about 5 to 10 minutes before the accident flight indicated that the NTSB Aircraft Accident Report 46 cloud ceiling over the south end of Ella Lake was between 1,400 and 1,600 ft msl (which is about 1,150 to 1,350 ft above the surface of the lake and 800 to 1,000 ft over terrain located south of the lake) and the visibility was slightly restricted by rain but was otherwise good. The Promech pilot whose flight followed the short route about 3 to 4 nm behind the accident flight described that conditions deteriorated as his flight approached the southern portion of the lake, with visibility less than 2 miles in rain and low clouds obscuring mountainous terrain. Based on all the available weather information, the NTSB concludes that the accident flight encountered deteriorating weather conditions over the southern half of Ella Lake, and, at the time of the accident, the terrain at the accident site was likely obscured by overcast clouds with visibility restricted in rain and mist. 2.2.2 Pilot’s Flight Route Deviation Despite the deteriorating weather conditions in the Ella Lake area, the two other Promech pilots on the short route (one who departed before the accident flight and one who departed after) successfully completed their trips. In reduced visibility weather conditions, Promech pilots typically fly southwest over Ella Lake and proceed beyond the south end of the lake before turning west. This route allows the flights to avoid the high terrain located west of the lake, which includes two similarly shaped mountains. (See figure 16.) Figure 16. The typical flight route (white) passes two similarly shaped mountains before turning west. For comparison, the path flown by the accident pilot is shown in green.

[ANALYSIS] Effective CFIT avoidance requires assessment of weather conditions and appropriate aeronautical decision-making in response to cues associated with deteriorating weather. The pilot missed opportunities to prevent the accident both in his decision to take the short route despite the presence of low clouds and obscured terrain, to turn toward an area of the lake where visibility was further reduced, and to continue the flight into deteriorating conditions rather than escape them. Research on weather-related accidents suggests that expert and novice pilots assess weather differently and that some weather-related accidents may result, in part, from an inaccurate assessment of weather conditions (Wiggins and O’Hare 2003a). The FAA defines (in 14 CFR Part 1) daytime flight visibility as the average forward horizontal distance from the cockpit at which prominent unlighted objects may be seen and identified; additional FAA guidance (in FAA Order JO 7900.5D) states that, if a reference marker “can barely be seen and identified, the visibility is about the same as the distance to the marker.” However, several factors can affect the extent to which landmarks are identifiable or barely visible to a pilot flying in poor visibility conditions, including the pilot’s contrast sensitivity, familiarity with the landmarks involved, and accuracy of distance estimates to those landmarks. Contrast sensitivity declines with age, beginning about age 40 (the pilot’s age was 64); therefore, the accident pilot may have had less ability to identify features in reduced contrast visual scenes like those obscured by rain or mist. Although the accident pilot was experienced, he had NTSB Aircraft Accident Report 49 relatively little experience flying tours around Ketchikan; thus, he likely had less finely detailed knowledge of the local geography. This same lack of knowledge of the fine details of the local geography could also have made it more difficult for him to estimate distance. Therefore, the pilot’s relative unfamiliarity with the landmarks on the tour route and the area weather dynamics may have affected his ability to accurately assess and respond to the deteriorating weather conditions he encountered. 2.2.3.1 Pilot’s CFIT-Avoidance Training and Experience The FAA does not require CFIT-avoidance training for pilots of fixed-wing aircraft. However, since 2012, all commercial air tour operators in Southeast Alaska have been providing cue-based training to their pilots that was developed specifically for their operations as a result of NTSB Safety Recommendation A-08-61, which recommended a cue-based training program that “specifically addresses hazardous aspects of local weather phenomena and in-flight decision-making.” Such training programs are based on the premise that exposing pilots to realistic depictions of deteriorating in-flight weather will help calibrate their weather assessment and foster an ability to accurately assess and respond appropriately to cues associated with deteriorating weather (Wiggins and O’Hare 2003b). Promech’s CFIT Avoidance Manual, which was created in conjunction with the Medallion Foundation, outlined its training program and described the policies and procedures for company pilots to use for CFIT avoidance during all phases of flight. Per the manual, CFIT-avoidance training included ground training and cue-based training in an FAA-approved ATD.

[ANALYSIS] He also stated that, during initial training, the assistant chief pilot told him and a group of other Promech pilots that they had to bend the rules because they were operating in Alaska, and that, if one pilot turned around while the others made it through, they were going to “have a conversation.” He said that, because he was new to flying commercially in Alaska, he assumed that this was the culture and that they were to push through the weather. Some of the company’s more experienced pilots said that they did not feel pressured to fly in unsafe conditions; however, one of these pilots was part of the group that encountered the 200-ft ceiling during the second tour of the day, and this pilot told investigators that he had never needed to turn down a flight for safety-related reasons. Another experienced Promech pilot said that he was ridiculed and threatened with termination for asking on the radio if another pilot (the accident pilot) was operating in IMC. 39 A National Aeronautics and Space Administration-sponsored study of terminal area thunderstorm penetrations by commercial airline flights found, for example, that pilots were more likely to penetrate convective weather if they were following another aircraft, they were behind schedule by more than 15 minutes, or they were flying after dark. Thus, the weather-related decision-making of pilots is influenced by other pilots (Rhoda and Pawlak 1999). NTSB Aircraft Accident Report 55 Considered together, these multiple accounts of pilots, including company managers, suggest the normalization of flying in weather conditions below FAA minimums. The NTSB concludes that, as evidenced by the company president/CEO’s own tour flights on the day of the accident, Promech management fostered a company culture that tacitly endorsed operating in weather conditions that were below FAA minimums. The DO provided an example of having praised the accident pilot on one occasion when he was the only pilot in one particular group of flights to abort a tour and return to base because he felt the weather conditions were unsafe; one of the other pilots who had flown in that group said that the other pilots also reacted positively to the accident pilot’s decision. Although this superficially suggests a safety-oriented organizational culture, there is no indication that Promech questioned the appropriateness of the other pilots’ decisions to continue their tours in the same weather conditions. As a result, it is likely the pilot’s decision to return to base in adverse weather conditions was actually negatively reinforced because the actions of the other pilots indicated that it was acceptable to continue. The risky behavior exhibited by other Promech pilots and management may have increased the difficulty that the accident pilot had in evaluating the risks associated with adverse weather, particularly considering that he had relatively little experience flying in the Ketchikan area. Research has identified that difficulty evaluating risks is an important factor in weather-related accidents (Johnson and Wiegmann 2015). The accident pilot’s more conservative decision-making when flying alone versus his continued flights into adverse weather when following other Promech flights (including those piloted by company managers and another experienced pilot he admired) provides evidence that the accident pilot struggled with calibrating his own perceptions of risk in Promech’s risk-tolerant organizational culture before the accident.

[ANALYSIS] Therefore, the NTSB recommends that the FAA implement ways to provide effective TAWS protections while mitigating nuisance alerts for 36 For more information, see NTSB case number ANC14MA008. 37 For more information, see NTSB case number ANC15FA049. NTSB Aircraft Accident Report 52 single-engine airplanes operated under Part 135 that frequently operate at altitudes below their respective TAWS class design alerting threshold. 2.2.3.3 Terrain Depiction: Chelton EFIS Software and Database Considerations The Chelton EFIS user manual includes procedures for using the system when escaping inadvertent VFR flight encounters with IMC. These procedures include reversing course, looking at the moving map to determine the direction to turn away from terrain, and (in the event of a terrain alert) identifying the threatening terrain on the moving map and maneuvering to avoid it. However, the original 2003 terrain database used in the accident airplane and older versions of the EFIS software (such as that installed in other airplanes) can limit the fidelity of the terrain information available to system users. The legacy Chelton system’s original 2003 database (which was installed in the accident airplane) does not distinguish small, inland bodies of water in blue on the terrain moving map but rather shows the area in the same color as the surrounding terrain; the 2007 terrain database, which is more detailed, depicts small, inland bodies of water (such as Ella Lake) in blue. The more-detailed updated database would have provided the accident pilot with better information for situational awareness near Ella Lake. More detailed information would also be useful to any pilot when performing the prescribed procedures for using the system when escaping inadvertent VFR flight encounters with IMC. As described above, the Chelton MFD depicts potentially hazardous proximity to terrain (based on the TAWS requirements) as a red (warning) or yellow (caution) overlay on the terrain map. Although these overlays have some degree of transparency in version 6.0B of the EFIS software (which was installed in the accident airplane) to allow the underlying contours to be distinguished to aid in a terrain avoidance maneuver, some pilots reported that the overlays in their airplanes could, at times, obscure the terrain depiction.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] Under these conditions the radar controller may have difficulty in observing the target presentation. However, in the case of Flight 45, the final controller stated chat he observed precipitation echoes on hz PAR scope between 1 and ? miles trom touchdown but that these echoes did not interfere with the depiction of the aircraft target. The final approach controlle: was several seconds late in transinitting target elevation (alsitude) display information. relating to the declaration of minimum altitude. That transmission was made when the aircraft was at an actual altitude of about 85 feet mus... as shown on the Right recorder, instead of when the aircraft was at the U.S. Air Force PAR mininoom altitude of 214 tect ms.b. The Board believes the final approach control‘er devoted continuing attention to the azimuth displays as evidenced by the nunierous heading changes given. This continuing attention to the azimuth display might have limited the atcention that he could devote to elevation observation. This situation combined with the increased rate of descent of the aircraft during the last 8 to 10 seconds of flight, could have - §y at if S t q 4 “ F 4 contributed significantly to the delay ia transmission of altitude informacion. The controllec’s cali, placing the aircraft et minimum altitude and too low to complete a safe approach, was not broadcast in sufficient time to aiert the pilot to his dangerously low position, Additionally, the sound of the rain removal equipmen: might have interfered with the tlightcrew's ceception of the controiler's calls. Thus, under the circumstances the warning effects associated with the controller’s minimum altitude call were negated and the firsc officer's impression chat everything was “OK” until they hit is quite understandable. Another factor which compounded the erew’s problems during the final descent was thar the tain removal system was not operating at total capacity in removing the water from the windshield. The reduction in capacity was due to low engine r.p.m. during the last 22 secends of flight. Thus, the accumulation of raindrops, with associated problems of refraction or distortion and possidle depressed horizon, limited the crew’s efforts to see the runway during the most critical portion of the approach. Subsequent to the level-off, the crew probably expected to break c--¢ or to obtain visual contact with the runway .ghts momentarily. The power reduction, the position in the rain shower and short approach light system could have contributed to a delay in their obtaining visual contact during which time they got into an unperceived high rate of descent from which there was ito recovery. The Board considered the possibility that erroneous barometric altitude information misled the crew during passage through the rain shower on final approach. Information concerning pressure and wind changes that occur in chunderstorms in the middie western portion of the United States was reviewed in an effort to associate the changes with those that existed at Naha at the time of the accident. However, there is nothing in the FDR trace to suggest that conditions similar to those observed in midwestern thunderstotms existed in the rain shower.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] The combination of wind change, application of minimal control forces, and reduction of power is the most plausible explanation for the high rate of descent prict ta impact. 2.2 Conclusions fa) Findings 1. The crew was trated. certificated. and qualified in accordance with existing regulations. 2. The aircraft was certificated in accordance with existing Federal Aviation Regulations and had been maintained in accordance with existing FAA and Flying Tiger Line, Inc., directives. 3. The aircraft was airworthy and there was no evidence of mechanical failure. 4. Flying Viger Line, Ine., dispatch procedures were in accordance with applicable regulations. “ ce a in a7 wr a va et ay nee OE 5. The aircrafe weight and balance were within limits, 6. Flying Tiger Lire, Inc., precision approach radar minimums for Naha Air Base were 300 feet and 3/4-mile visibiliry. 7, Vhe approach was flown to DH in accordance with the final controller's instructions, 8. The first officer flew the approach tc DH from the right seat. 9, Ther: was a heavy rain shower in the vicinizy, approximarely 1 mile in diameter, at minimum descent altitude, The area surrounding the rain shower was brightly lighted by midday sunlight. 10. A 1€ to 15 knot tailwind from 120° existed on the final approach. 11. The efficiency of the pneurnatic air rain remaval system was reduced by low power setting during the final pottion of the approach. 12, The required approach altitude calls were not made. 13. The flight operating procedures in the Operations Manual were not spec ific with tegard to altitude -allouts by the captain and first officer, when the latter was flying the aircraft ov an instrument approach. 14, The captain’s radio altimeter veference bug was set improperly at 200 feet, the first officer's was set correctly at 300 feet, and the altitude reminder dial was set at 310 feet, 13. The aircraft leveled at about 325 fect msl, and povicr was reduced. Powss was never increased thereafter, 16, Correlation of the CVR and FDR information showed that someone in the cockpit called out Hundred feet” at the same time the aircraft was at 100 feet m.s.1, 17. No action was taken by cither pilot when chis call was made. 18. The final approach controller advised the crew that they were at minimums (200 feet) t-second after the “hundred feet" call in the cock pit, 19.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] Simulator studies were conducted by the Board to examine the last portion of the approach. These sudies showed chat if the 15 knots tailwind ceased for 10 secands when the aircraft passed through about 400 feet it would result in a flight recorder trace similar to the one obtained from the accident. Additionally, the reduction in power to an Ny compressor speed of 60 percent is a natural result of observing an airspeed 10 knots higher than desired, Witnesses varied in their evaluation of the init'al impace attitude of the aircraft. One witness stated that the aircraft struck the waver in a noscup attivude, whercas other witnesses described the aircre“e striking the water ia a nosedown attitude, This variance in witness obseivetions can be reconciled by the difference in visual angles, distance from the impact area, and the reduced visibility caused by the rain shower in the area of the accident. The crew adhered closely to hzading and altitude instructions until level off upon reaching Flying Tiger minimums (300 feet), Thereafter, however, the aircraft descended through the Decision Heignt and contacted che water. Accordingly, the investigation was directed toward determining what factors may have led to this unwarranted descent. The Safety Board has considered and ruled out the following as possible mechanisms of causation: (1) In-flight failure, malfunction, or abnor- mality that would have caused or contributed to an unwarranted rate of descent. (2) An unauthorized person in the cockpit. (3) Pilot fatigue. (4, in-flight pilot incapacitation. The crewmembe:s were all performing their duties and conversiug in normal tones until just before the accident a-curred. i: was determined fram the CVR that the first officer flew the final approach ro Naha while the captain handled the communications and maintained external reference. As the aircraft progressed into the rain shower, the crew probably lost excernal visual reference completely, dve to the intensity of the rain, However, since the flight was nearing approach minimums and they expected to break out of the shower momentarily, the captaia undoubtedly devoted his attention to locating the approach lights. Durirg the 7-second period of level flight at 325 feet ms.i, the aircraft passed through the most intense portion of the shower and emerged into an area of increased light intensity. The bucklighted light rain could well have caused visual disorientation effects associated with an il’uminated high intensity “Ganzficld” phenom. enon (a homogencous visual field of similar brightness in which no differentiating objects can be seen). The give no: only would preclude reference to outside objects but also would limit reference io cockpit instruments. During this period the final approach controller advised the crew that they were ‘“‘on glidepath” and had a 10 knot tailwind.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] One possible explanation for the lack of those calls is that the flight was in visual conditions at both times, and consequently, the pilots may not have regarded the calls necessary. Another explanation could be the lack of explicit written company directives as to crew duties when the first officer is flying the aircraft on an instrument approach. Also, there was an evident lack of a clear understanding between the pilots as to what the DH should be. The first ofticer determined correctly that the DH was 300 feet (314 feet m.s.Lj as his radio altimeter bug was found set at 300 feet; he had leveled the aircraft at about 325 ausd. for a period of 7 seconds; and the altitude reminder dial was found set at 310 feet. The captain apparently had a minimum altitude of 200 feet in mind. His radio and pressure altimeter bugs were found set at 200 feet, and he was certified (under certain circumstances not existing in this case) as qualified for the lower minima. Assuming that the captain intended to make the “100 feet above minimums” call, a normally accepted procedure, he could not have been expected to make the call since the aircraft did not reach the 100-foot point (314 feet m.s..) above his 200 feet minimuins until the aicraft departed the lovel-off altitude of 325 feet. During the reinitiated descent, his attention probably was devoted to locating the runway. The correlation of the CVR and FDR information showed that the aircraft started the descent approximately 1/2-mile before it reached the poine where the published glidepath intercepted the 1,000-foot level and that it remained consistently below the glidepath until the level-off at about 325 feet. This condition prevailed even though the aircraft was reported by the GCA final controller to be on the plicdepath. The Board examined the precision approach radar unit for various defects which could have existed ana could cause the aircrafs to appear to be positianed differently on the glidepath from the one observed by the controller. ‘The results of this examination were inconclusive as to the type of defect that could have caused a below-glidepath condition, since any such defect would show up on the radarscone and alert the controiler to a problem with the equipment. Additionally, taining procedures include alerting the controler to any abnormal radarscope presentation. The Board recognizes that under certain conditions water drops have an etfect on the radarscope presentatioa in heavy concentrativns of water. Under these conditions the radar controller may have difficulty in observing the target presentation. However, in the case of Flight 45, the final controller stated chat he observed precipitation echoes on hz PAR scope between 1 and ? miles trom touchdown but that these echoes did not interfere with the depiction of the aircraft target. The final approach controlle: was several seconds late in transinitting target elevation (alsitude) display information. relating to the declaration of minimum altitude.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] At the time radar contact was established (1129) the flight was 18 miles ncrehwest of the airport and was apparently operating in visual neteorelogical conditions, az they had dcen during virtually tiie entire flight from Tokyo. The crew was aware that the rain shower they observed in the vicinity of the airfield was local in nature. In addition, they were aware that either they would be making a downwind landing or they Would have to obeain an amended clearance if they chose to land into the wind, However, the crew made no effort tu circumnavigate the rain shower, by requesting a clearance to land into the wind. At 1129, che controller advised that there was construction equipinent on the left side of the ruaway at the approach end and on the righ, side of the runway at the 3,000 feet remaining marker. Two minutes later, at 1131, the controller advised further: ‘.. Jhave reduced visibility on final... tower just advised approach lights and strobe lights are on... .” Cockpit conversations reflect the crew's dis. pleasure with the location of reduced visibility due to skower activity. Thus following 2 hours of relative inactivity, the crew was faced with the necessity of executing an instrument approach with the attendant requirements for precise rapid responses to control instructions and environmental :ues, The FDR readout shows that a descent was established on the glidepath at a rate of about 950 fip.m. at a point slightly less than 5 miles from the runway. The descent rate was reduced to 750 f.p.m, about 3 miles from the runway, The crew activated the pneumatic rain removal sy stem about 7 seconds prior to the final controller's “two miles..." transmission. At this time. the rate of descent increased to about 930 f pan. and continued at that rate until com. inencement of a level off about 10 seconds later. Completion of the level-off maneuver required about 4 seconds at 325 feet ms.l. or about this crew's decision height. The airspeed increased from: 153 knots to 163 knots during the descent within the rain shower, During the level Might at 325 feet, the power wis reduced co an NI compressor speed of apprortmately 60 fercent r.pam. and the airspeed, thereupon, was reduced to 154 knots. This speed remained constant during the last 22 seconds of flight ;5 seconds of level flight followed by 17 seconds of descent! at an average rate of 1,150 fect per minute. Simulator studies were conducted by the Board to examine the last portion of the approach. These sudies showed chat if the 15 knots tailwind ceased for 10 secands when the aircraft passed through about 400 feet it would result in a flight recorder trace similar to the one obtained from the accident.

[CONCLUSIONS] No action was taken by cither pilot when chis call was made. 18. The final approach controller advised the crew that they were at minimums (200 feet) t-second after the “hundred feet" call in the cock pit, 19. The sircralt was seen emerging from the rain shower 78 to 100 feet above the water, 20. The areraft contacted the water 2.3 seconds after the beginning of the “at minimums” call by the final approach controller, 21. The final approach simulation showed that the FDR traces could be approximated by: (1) programming the known pres. sure and vemiperature conditions; (4) assuming a 15 knot tailwind and removing «he taiiwind at 450 feet Cor 10 seconds; and then reinserting it; (3) Reducing the sower to 60 percrnt Ny when an inctcase in airspeed aud level off was noted, and then applying only a minimum amount of cotttrol force. 22, Meteorolcgical conditions created a veiling glare and a visual field of similar brightness. 23, No evidence of malfunction of the static system instruments was found. 2A, There was no evidence of the existence of meteorological conditions severe enough to cause the altimeter to read in error. 25. The accident was survivable; however, the captain was not wearing his shoulder harness and died asa resule of injuries, 26. The aircraft was destroyed by impact and there was no fire. 27. Of the four crewmembers on board one died as a result of injuries, one from asphyxiation and two from drowning. (b}} Probable Cause The National Transportation Safety Board determines that the probable cause of thir accident was an unarrested rate of descent duc to inattention of the crew to instrument ale:tude references while the pilot was atrempting to establish outside visua! contact in meteorological conditions which precluded such contact during that seginent of a precision radar approach intound from the Decision Height. 3. RECOMMENDATIONS During the latter part of 1968 and the early part of 1969, arash of serious aircraft accidents occurred during the instrument approach phase of flight. As a results of those accidents, the National Transportation Safety Board. by letter dated January 17, 1969, made a number of recommendations to the Administrator of the Federal Aviation Administration, Among others, it recommended that the Administrator: emphasize the importance of altwude awareness during instrument approaches through strict attention to instrument indications, ceew coordination, and altitude callout procedures.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] During this period the final approach controller advised the crew that they were ‘“‘on glidepath” and had a 10 knot tailwind. The “on glidepath” transmission undoubtedly reassured the crew regarding their altitude and position. About 8 seconds «fer this ‘ton glidepath” portion of the controller's transmission, an unidentified person in the cockpit (probably the second officer) called out “hundred fect.” At that point the flight recorder indicated an altitude of 85 feet m.s.. There was no evidence that this call alarmed the captain or the first officer, even though the radar aitimeters would have indicated the alticude. The an:ber warning I'ghts associated with these radar altimeter: would have been lit also, since the captain’s reference “bug” was set at 200 feet m.s.l. and the first officer's was set at 300 feet m.s.l. One-second after the “hundred feet" call the controller advised the flight: ‘‘at minimum altitude, going well below glidepath, too low...” The CYR recording of the controller's instre.- tions ended at that point. While the final a¢- proach controller was making that transmission, the unidentified person in the cockpit was calling, “Seventy feet" and “It's fifty fer.” If the laccer calls alarmed the captain and/or the first officer, it was too late to refocus on the instruments, interprec them, and effect a recovery from the relatively high rate of descent chat existed during the last few seconds of flight. During the period from 1134:14 (5 miles from touchdown) to 1435:37 (1 nule from touchdown and 8.5 seconds prior to impact), the contro'hr gave the flight six heading changes. The first officer was required to concentrate on making directional changes which may have precluded an effective overall surveillance of other instruments, particularly the aleim. eter. Thus, he may have been relying on the controilet to provide altitude information. There was ao evidence that either the “500 feet above minimums” or ‘100 feet above mini. imems’’ call, required by company directives, was made. One possible explanation for the lack of those calls is that the flight was in visual conditions at both times, and consequently, the pilots may not have regarded the calls necessary. Another explanation could be the lack of explicit written company directives as to crew duties when the first officer is flying the aircraft on an instrument approach. Also, there was an evident lack of a clear understanding between the pilots as to what the DH should be.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] Information concerning pressure and wind changes that occur in chunderstorms in the middie western portion of the United States was reviewed in an effort to associate the changes with those that existed at Naha at the time of the accident. However, there is nothing in the FDR trace to suggest that conditions similar to those observed in midwestern thunderstotms existed in the rain shower. Consideration was given to the possibility that Flight 14 45 might have encountered severe up or down drafts during passage through the rain shower, buc the FPR trace shows no indication of such an occurence, Furthermore, the U. S. Air Force C-130 pilot who had complete an approach shortly before Flight 45 began its approach, did not encounter severe condition’ ‘vithin the cain shower. The Board.’s +.-Iisdful of the rapid and marked surface pressuie ariatious which usually occur in a particular sequence cl.aracterized by: (1) Falling pressure as the storin approaches; (2}} An abrupe rise in pressure associated with rain showers as the storm moves overhead; and, (2) A gradual return to nurmal pressure as the storm moves on and the rain ceases. Thus, atmospheric pressure within an area of heavy precipitation as at Okinawa would be slightly higher than the pressure in the surrounding environment. Flying towards a zone of higher pressure, at a particular flight level, and with a constant altimeter setting, causes the altimeter to read too low. The indicated altitude is lower than the actual altitude. Accordingly, as in the case of Flight 45, the error, if any, would have been on the safe side. In view of the above, the Board finds no evidence to indicate that atmospheric pressure fluctuations were involved in the causal area of this accident. : A pressure difference of approximately 33 inches of mercury would be required besween the ambient air pressure and the static system pressure to obtain a 300-foot altimeter error. The airspeed indicator would read coacomitantly about 23 knots in error (high in this case). The possibility was considered that water ingestion in the static pressure system ports might have caused altimetry errors which led the pilots to believe they were approaching decision height when in fact they had descended through EEE YEE RTRSY Be aE ee een ee e. a4 ° it, Tests conducted on static systems of other aircraft Lave shown thar both altimeter and airspecd indicators experience noticeable excursions when water is being ingested. Therefore, a £00 altimeter error due to water iriyestion should have been obvious to the crew. The alternate static port isthe source of stati: system pressure for the flight recorder. The normal source, located aft of the alternate source, provides pres.ire for operation of the primary instruments. Phung calibration and certification tests both rain and icing conditions were experienced, The static pressure difference between the normal and alternate systems in the aircraft configuration here being examined is 17 feet of altitude and 1 knot of airspeed.

[ANALYSIS] At the time he received the clearance, he was about 44 nmi from Armel on the 300° radial inbound to the station. By reference to the approach chart, he should also have been able to identify the high obstacles between that position and the. Round Hill intersection. With that information, he should have been able to determine that 1, 800 feet was not an adequate altitude to provide terrain clearance of 2,000 feet in this designated mountainous area. If he did not realize that he was over a designated mountainous area, he should have applied terrain clearance of 1,000 feet as prescribed for nonmountainous areas. He did notice the 3,400 feet associated with the course between Front Royal and Round Hill. That should have suggested that he should reexamine his decision regarding the descent to 1,800 feet. If he had questioned the controller regarding the minimum altitude in the area of his aircraft, he should have received information that would have alerted him that he could not descend to 1,800 feet until after he passed Round Hill. The information available to the pilot, including the approach chart, should have alerted the crew that an unrestricted descent would be unsafe. It does appear to the Board that there was a deficiency in the chart. This particular approach chart depicted the profile view from the final approach fix to the airport. It did not depict the intermediate fix, Round Hill, with its associated minimum altitudes. This information was available from the plan view of the chart, but it appears that the crew gave their primary attention to the profile. If this was the case, it may have led the crew to discount the other information available on the chart and to continue their descent on the assumption that it was permissible by reason of the clearance they received. The second major question deserving consideration is the role I of the ATC system inthis accident, specifically why TWA 514 was not given an altitude restriction in its approach clearance. The testimony of all FAA witnesses, including the controller, was consistent in stating that Flight 514 was not a "radar arrival;"' that because of this fact the controller was not required to implement the provisions of paragraph 1360 of the FAA Handbook 7110. 8C; and that they considered TWA 514, after intercepting the 300° radial of Armel, as proceeding on its own navigation and as being responsible for its own obstacle clearance. The FAA witnesses stated that Flight 514 was not a radar arrival because it had not been vectored to the final approach course. They did not consider the vector of Flight 514 by the Washington Center to intercept the 300° radial as being a vector to the final approach course, even though the VOR/DME approach procedure utilizes the 300° radial inbound from Round Hill. Particular emphasis was made by FAA that the vector to the 300° radial occurred when the flight was approximately 80 miles from the airport and that it was vectored by the center on to an en route course.

[ANALYSIS] This indicates that the altimeter system was operating properly. The elevation at impact was about 1,675 feet. The altimeter was set at 29.70, the last altimeter setting given to the crew. Two reasons why the aircraft might have been below its target: altitude of 1,800 feet are evident. First, the aircraft was entering I ground effect as it got closer to the ground and this may have caused an error in the pitot static system which caused the altimeter to indicate an altitude higher than the actual aircraft altitude. Second, it is possible that the high winds blowing over the rough terrain in the accident area may have caused a pressure change which affected the altimeter indication. However, the crew's evident concern about the altitude was indicated by the captain's order regarding the power and the first officer's comments about the downdraft when the aircraft . ‘went below the target altitude. Based on the evidence available, the Safety Board concludes that there was no significant error in the altitude information presented to the pilots by their instruments. The crew's comments regarding the altitude and the power indicate that the first officer was not flying the aircraft at the target altitude of 1,800 feet. The Board examined the flight data recorder trace and found that while there was evidence of moderate turbulence, it was probably not of sufficient magnitude to prevent the first officer from maintaining the desired altitude. There was also no evidence that there was any problem within the aircraft that would have prevented the pilot from staying at 1,800 feet. Therefore, the Board concludes that the deviation below the target altitude was probably a result of the combination of the first officer's flying technique and the turbulence. From the above, it is clear that this was an operational accident and that the crew knowingly descended to approximately 1, 800 feet after being cleared for the approach. The basic questions requiring resolution are (1) why did the crew knowingly descend to 1,800 feet in an area where the terrain obstacles extended almost up to that altitude; and (2) why did the approach clearance not include an altitude restriction under the circumstances of this case. Our review of the record supports the conclusion that the captain believed that when he approached the airport in a radar environment for a nonprecision approach he would not be "cleared for the approach" without an altitude restriction unless he could make ~ 28 -~ an unrestricted descent to the final approach fix altitude. In attempting to determine the reasons for the captain's belief in this regard, a brief description of the development of the usage of radar and its impact on pilot responsibilities is required. Before the advent of radar, the pilot alone was responsible at all times for knowing the position of his aircraft with regard to the terrain. The pilot kept the controller informed of the aircraft's position and of the pilot's intentions. Typically, during an instrument approach, numerous radio calls were made as the pilot reported his position, altitude, and intentions. With the advent of radar, the controller was able to observe the aircraft in two dimensions -- range and azimuth -- and was able to _ vector flights to arrive over geographical positions.

[CONCLUSIONS] The first officer's altimeter was set properly. It is possible that wind velocity over the hilly terrain may have induced an altimeter error which could have caused the instrument to indicate that the aircraft was higher than its actual altitude. However, the crew's last.comments regarding altitude indicated that they knew they were below 1,800 feet. The altitude alerting system and the radio altimeter aural warnings sounded at appropriate altitudes to indicate to the pilots that the aircraft was below 1, 800 feet and that the aircraft was within 500 feet and 100 feet of the ground. These latter warnings occurred 7 seconds and 1 second, respectively, before impact. The flightcrew apparently did not have sufficient time to avoid the accident after these warnings. 13. 14, 15... 16. 17. 18. 19. The approach clearance was given to the flight without altitude restrictions because the flight was not being handled as a radar arrival and because the controller expected the crew to conduct the approach as it was depicted on the approach chart. Procedures contained in FAA's Terminal Air Traffic Control Handbook were not clear and resulted in the classification and handling of TWA 514 as a 'nonradar'' arrival. The terms ''radar arrival" and "nonradar. arrival'' were not defined. In view of the available ATC facilities and services and since the flight was receiving radar service in the form of radar monitoring while under the jurisdiction of a radar approach control facility, the procedure should have provided for giving altitude restrictions in an approach clearance for an aircraft operating on an unpublished route prior to its entering a segment of the published approach procedure. The ATC system was deficient in that the procedures were not clear as to the services the controllers were to provide under the circumstances of this flight. The flightcrew believed that the controller would not clear them for an approach until they were clear of all obstructions. The depiction on the profile view of the approach charts neither indicated the position of Round Hill intersection nor did it contain all minimum altitudes associated with the approach procedure. This information was available on the plan view of the approach chart. The captain noticed the minimum altitude associated with the approach segment from Front Royal to Round Hill but he decided that the flight could descend to 1,800 feet without regard for the 3, 400-foot minimum altitude depicted on the chart because he was not on that segment. 20. 21. 22. 23, 24. ~39- The captain of Flight 514 did not question the controller after receiving the approach clearance, regarding the action the flightcrew was expected to take. Another crew that questioned a similar clearance received further instructions and information which resulted in their accepting a radar surveillance approach to Dulles. ) Both military and civil aviation officials for several years had indicated concern regarding a lack of understanding on their part of what the Air Traffic Control procedures and terminology were intended to convey to the pilots.

[CONCLUSIONS] Another crew that questioned a similar clearance received further instructions and information which resulted in their accepting a radar surveillance approach to Dulles. ) Both military and civil aviation officials for several years had indicated concern regarding a lack of understanding on their part of what the Air Traffic Control procedures and terminology were intended to convey to the pilots. They were also concerned about the possibility of misunderstandings which could result in pilots descending prematurely. ' The FAA was not responsive to the long standing, expressed needs and concerns of the users of the Air Traffic Control System with regard to pilot/ controller responsibilities pursuant to the issuance of an approach clearance for a nonprecision approach. Furthermore, the FAA did not provide users of the Air Traffic Control System with sufficient information regarding the services provided by the system under specific conditions. The FAA did not utilize the capability of the ARTS III system to insure terrain clearance for descending aircraft conducting nonprecision instrument approaches in instrument meteorological conditions. The flightcrew of Flight 514 was not familiar with the terrain west and northwest of Dulles. However, they did have information regarding the elevation of obstacles west of Round Hill intersection depicted on the plan view of the approach procedure. b.. Probable Cause The National Transportation Safety Board determines that the probable cause of the accident was the crew's decision to descend to 1, 800 feet before the aircraft had reached the approach segment where that minimum altitude applied. The crew's decision to descend was a result of inadequacies and lack of clarity in the air traffic control procedures which led to a misunderstanding on the part of the pilots and of the controllers regarding each other's responsibilities during operations in terminal areas under instrument meteorological conditions. Nevertheless, the examination of the plan view of the approach chart should have disclosed to the captain that a minimum altitude of 1, 800 feet was not a safe altitude. ‘ , Contributing factors were: (1) The failure of the FAA to take timely action to resolve the confusion and misinterpretation of air traffic terminology although the Agency had been aware of the problem for several years; (2) The issuance of the approach clearance when the flight was 44 miles from the airport on an unpublished route without clearly defined minimum altitudes; and (3) Inadequate depiction of altitude restrictions on the profile view of the approach chart for the VOR/DME approach to runway 12 at Dulles International Airport. 3, RECOMMENDATIONS As a result of the accident, the Safety Board submitted 14 recommendations to the Administrator of the Federal Aviation Administration. (See Appendix I.) Subsequent to the accident, the FAA has taken several actions in an effort to prevent recurrence of this type of accident. 1. The FAA has directed that all air carrier aircraft be equipped with a ground proximity warning system by December 1975. 2. The FAA has revised the provisions of 14 CFR 91 with regard to pilot responsibilities and actions after receiving a clearance for a nonprecision approach. 3. The FAA has established an incident reporting system which is intended to identify unsafe operating conditions in order that they can be corrected before an accident occurs. .~ 4] - 4. The FAA has changed its air traffic control procedures to provide for the issuance of altitude restrictions during nonprecision instrument approaches. 5. The FAA is installing a modification to the ARTS III system that will alert air traffic controllers when aircraft deviate from predetermined altitudes while operating in the terminal area.

[ANALYSIS] The possible effect of the high winds on the indicated altitude has been discussed previously. While the evidence does not indicate whether the crew was aware of the SIGMETS issued for the Washington area, there is no evidence to indicate that knowledge of the SIGMETS would have caused the crew to operate any differently than they did. "= 35 - The CVR indicates that the crew did encounter considerable turbulence during the descent. However, the record also indicates that they.were able to read the altimeters well enough to know that they had descended below their target altitude of 1,800 feet. The Safety Board believes that the effect of turbulence was not critical but could not determine positively why the descent was not arrested at 1,800 feet. . In summary, this accident resulted from a combination of . conditions which included a lack of understanding between the controller and the pilot as to which air traffic control criteria were being applied to the flight while it was operating in instrument meteorological conditions in the terminal area. Neither the pilot nor the controller understood what the other was thinking or planning when the approach clearance was issued. The captain did not react correctly to his own doubt about the line of action he had selected because he did not contact the controller for clarification. The action of the other air carrier pilot who questioned the clearance he received about 1/2 hour before the accident is the kind of reaction that should be expected of a pilot suddenly confronted with uncertainty about the altitude at which he should operate his aircraft. The Board again stresses that it is incumbent upon air carrier management to assure the highest possible degree of safety through an assertive exercise of its operational control responsibility. This management function must assure that flightcrews are provided with all information essential to the safe conduct of flight operations. Furthermore, the air carrier must assure that its flightcrews are indoctrinated in the operational control precept and that during flight the final and absolute responsibility for the safe conduct of the flight rests solely with the captain as pilot-in-command regardless of mitigating influences which may appear to dilute or derogate this authority. Whereas the air carriers and the pilots are expected to per-— form their services with the highest degree of care and safety, this same high level of performance must be expected from the management of the air traffic control system and the controller. The instant case provides a classic and tragic example of a pilot and controller who did not fully comprehend the seriousness of the issuance and’ acceptance of a clearance which was not precise or definitive. The pilot should question a clearance which leaves any doubt as to what - 36 -. 13/ course of action should be followed. —- The Board also believes that it is incumbent upon the controller to ascertain beyond a doubt that the terminology of a clearance conveys the intent to the pilot, and to question the pilot if there is any doubt that he has understood it and is initiating actions compatible with the intent of the clearance. --- Footnotes: [13/ course of action should be followed. —- The Board also believes that it is incumbent upon the controller to ascertain beyond a doubt that the terminology of a clearance conveys the intent to the pilot,]

[CONCLUSIONS] During the course of the investigation it became clear that there was an omission in the ATC handbook concerning exactly when radar service is terminated. It was unfortunate that the handbook did not clearly ~ 44 - require the controller to provide altitude restrictions when an aircraft is operating over an unpublished route for which there is no minimum enroute altitude prescribed, while the flight was being handled as a non-radar arrival. The FAA has since limited such clearances and some action is being taken to correct the deficiencies cited above. “Yond A. Ss, Isabel A. Burges Member December 2, 1975 -~ 45 - McADAMS and HALEY, Members, dissenting: We do not agree with the probable cause as stated by the majority. In our opinion, the probable cause was the failure of the controller to issue altitude restrictions in accordance with the Terminal Air Traffic Control Handbook 7110. 8C, paragraph 1360(c), and the failure of the pilot to adhere to the minimum sector altitude as depicted on the approach plate or to request clarification of the clearance. Asa result, the pilot prematurely descended to 1, 800 feet, The flight was a radar arrival and, the refore, entitled to altitude protection and terrain clearance. If the controller, as required by the thenexisting procedures for radar arrivals, had issued altitude restrictions with the approach clearance or had deferred the clearance, the accident probably would not have occurred, On the other hand, if the pilot had either maintained the minimum sector altitude of 3, 300 feet as depicted on the approach plate, or requested clarification of the clearance, there would not have been an accident, ‘ The majority states (p. 32): "The Board concludes that based on the criteria in 7110. 8C the system allowed for the classification and handling of Flight 514 as a nonradar arrival. The Board, however, believes that the flight should have been classified and handled as a 'radar arrival,'"' This statement cannot be reconciled with the probable cause as stated by the majority. If the majority believes that under all tLe circum - stances the flight should have been classified and handled as a radar arrival, then the flight was in fact a radar arrival and the probable cause should so state. Itis not possible to determine from the majority opinion whether Flight 514 was a tadar or a nonradar arrival. The Board attributes the failure of the controller to handle the flight asa radar arrival to be a terminology difficulty between pilots and controllers. There was no terminology difficulty. .The plain fact of the matter is that the controller simply did not treat the flight as a radar arrival as he should have, All the criteria of paragraph 1360 for a radar arrival were present. Neither the pilot nor the controller had terminology difficulties. The pilot assumed he was a radar arrival and would be given altitude restrictions ifnecessary.

[ANALYSIS] Qualified instrument pilots and air traffic controllers should know and understand beyond equivocation that the coincidence of the inbound course being an extension of the final instrument approach course does not permit ‘descent to altitudes lower than those published for that air space segment unless specifically authorized by ATC. A clear, precise definition of final approach course and final instrument approach course should preclude future misunderstandings. Neither of these terms was defined in the AIM at the time of this accident. However, the AIM glossary did contain a definition of ''Final Approach - IFR" wherein the final instrument approach course is shown to be confined to the final approach segment of the instrument approach procedure and that it begins at the final approach fix. _ The issue of when flights are or are not radar arrivals must also be resolved. It is difficult for a pilot who is operating in a radar environment and communicating with a radar controller to realize that, under some circumstances, his flight is, without formal notification, considered to be a nonradar arrival and subject to a different ATC procedure. Specifically, he may not realize that the responsibility for obstacle clearance shifts from the controller to the pilot under some circumstances without the pilot being specifically informed. While the Safety Board recognizes that the FAA is concerned about radio frequency congestion in busy terminal areas, any control procedure which effects a change in the responsibility for providing terrain clearance must be communicated and clearly understood by both pilots and controllers. If radar service is terminated, the crew should be so informed. Then they will be prepared to resume the responsibility for navigation which was vested in the controller while the flight was classified and handled as a radar arrival. The ARTS III system provides, as previously noted, information capability not formerly available to controllers. The Safety Board has previously recommended that the altitude information capability of this equipment be used as an additional safety factor in the terminal area to help prevent controlled flight into the ground. In the case of Flight 514, the controller testified that he could not clearly see the target associated with the flight until he noted that the altitude was 2,000 feet. Immediately thereafter, he attempted to contact the flight to verify its altitude, but impact had already occurred. The FAA has taken action to install an altitude deviation warning in the ARTS III system which should be beneficial in alerting controllers to altitude deviations in the terminal area. Although the record of this investigation shows that the weather was a factor in the occurrence of the accident, it was not of such nature as to have made the accident inevitable. The icing encountered by the aircraft in the descent was apparently eliminated by the anti-icing systems. The intensity of the turbulence may have been sufficient to make the control of the aircraft somewhat difficult. The excursions of the traces on the flight data recorder are indicative of light to moderate turbulence. The possible effect of the high winds on the indicated altitude has been discussed previously. While the evidence does not indicate whether the crew was aware of the SIGMETS issued for the Washington area, there is no evidence to indicate that knowledge of the SIGMETS would have caused the crew to operate any differently than they did. "= 35 - The CVR indicates that the crew did encounter considerable turbulence during the descent.

[CONCLUSIONS] There was no terminology difficulty. .The plain fact of the matter is that the controller simply did not treat the flight as a radar arrival as he should have, All the criteria of paragraph 1360 for a radar arrival were present. Neither the pilot nor the controller had terminology difficulties. The pilot assumed he was a radar arrival and would be given altitude restrictions ifnecessary. Not having received such restrictions, he initiated a descent to 1, 800 feet. ~ 46 - Additionally, the Board concludes on the subject of radar arrival (p. 32): ",,.under these circumstances, [the clearance] should have included an altitude restriction until the aircraft had reached a segment of the published approach procedure or the issuance of the approach clearance should have been deferred until the flight reached such segment. . Therefore, the Safety Board concludes that the clearance was inadequate and its issuance and acceptance was the result of a misunderstanding between the pilot and the controller." Such a conclusion can again only mean that the flight was in fact a radar arrival since altitude restrictions are issued only in accordance with paragraph 1360(c), the provisions of which pertain solely to radar arrivals. Therefore, based upon the foregoing, it would appear the majority believes the flight was a radar arrival but refuses to make an unambiguous finding to that effect. . The Board further states (p. 32) that ''there is a general lack of understanding between pilots and controllers in their interpretations of air traffic control procedures.'' We find that there was no misunderstanding in this instance on the part of the pilot. As previously stated, he undoubtedly descended to 1, 800 feet after receiving an approach clearance because he was not issued an altitude restriction. If the controller was confused with regard to the application of paragraph 1360,he should have _asked for clarification from his supervisor. But there should have been no reason for confusion insofar as terminology is concerned. One of the most important functions of an air traffic controller is to possess the highest degree of knowledge in procedures and terminology and to apply it with the greatest diligence and care. . In any event, we can only conclude that, innot handling the flight as a radar arrival, the Dulles controller did not properly apply the provisions of the controller's handbook.

[ANALYSIS] Particular emphasis was made by FAA that the vector to the 300° radial occurred when the flight was approximately 80 miles from the airport and that it was vectored by the center on to an en route course. Operational advan_ tage was indicated by the controllers as the reason for the vector to the 300° radial rather than to an initial approach fix on the approach procedure. , The counterposition is that Flight 514 was operating ina .radar environment, was receiving at least one type of radar service, and was on a course which would lead directly to the Round Hill intermediate approach fix. Furthermore it had been advised that the reason for the vector to the 300° radial was for a VOR/DME approach for runway 12. Consequently, it should have received services, including altitude restrictions, as set forth in Paragraph 1360 of 7110. 8C. , In evaluating these facts, the one issue present is whether the handling of Flight 514 required the provision of an altitude restriction. FAA witnesses agreed that, had Flight 514 been classified as a radar arrival within the meaning of the handbook, the flight would have been given an altitude restriction until it reached Round Hill. In resolving this issue, the Board has been troubled by the fact that ATC procedures are almost always dependent upon the usage of certain specified phrases and terms, many of which have no established definitions and mean different things to controllers and pilots. The term "radar control''is an example. The pilot witnesses believed that, when they were operating in a traffic control radar environment, they were being controlled by radar. The controller group was aware that this was not always the case, but the FAA apparently did not perceive the difference of understanding, and the efforts made by the FAA to clarify when an aircraft was or was not radar controlled did not eliminate the confusion. The Board concludes that based on the criteria in 7110. 8C the system allowed for the classification and handling of Flight 514 as a nonradar arrival. The Board, however, believes that the flight should have been classified and handled as a "radar arrival." This, however, does not dispose of the issue of whether the ATC system should have provided for a redundancy that would have prevented or consequently identified and corrected a deviation of an aircraft from a clearance which was not followed as the controller expected it to be. The system should clearly require controllers to give the pilots specific information regarding their positions relative to the approach fix and a minimum altitude to which the flight could descend before arriving at that fix. Pilots should not be faced with the necessity of choosing from among several courses of action to comply with a clearance. . The Board believes that the clearance, under these circumstances, should have included an altitude restriction until the aircraft had reached a segment of the published approach procedure or the issuance of the approach clearance should have been deferred until the flight reached such segment.

[CONCLUSIONS] I knew that the aircraft would be vectored to the 300 degree radial. "Q. And how did you know that they would be vectored to the 300 degree radial? "A. Well, that's the final approach course." (Tr. 1106-07) "©. Well, prior to the implementation of the VOR-DME approach at Dulles, was there any training, or did you participate in any discussions concerning the conditions under which the 300 degree radial would be used? "A, [FAA witness, Dulles Supervisory ATC Specialist] No, there was no training or discussion. There -- it was on the approach plate as the final approach course. And the people had the approach plates. "Q. You're referring now to the NOS chart? "A. Yes, sir." (Tr. 1153) ~ 52 - 8. While there is no published definition of a final approach course, common usage over the years has extended that course outward with no mileage limitation as far as reasonable, depending on the usable reception i key of the facility. — The majority, however, has taken an ambiguous position on the most critical issue in the case -- was TWA 514 a radar or a nonradar arrival? Nevertheless, despite our conclusion that the flight was a radar arrival and therefore should have been provided altitude restrictions, the crew had at their disposal sufficient information which should have prompted them either to refrain from descending below the minimum sector altitude or, at the very least, to have requested clarification of the clearance. Although the profile on the approach plate did not fully and accurately depict the various minimum altitudes associated with the entire approach, it appears there was adequate information on the plan view of the plate to alert a prudent pilot of the hazards of descending to an altitude of 1, 800 feet prior to reaching the Round Hill intersection. The existing air traffic control system and today's aircraft are highly complex and sophisticated. Neither can operate independent of each other -- there must be a cooperative and coordinated effort on the part of both the pilots and the controllers if the system is to function efficiently and safely. The real issue in this accident is not one of inadequacy of terminology or lack of understanding between’controllers and pilots. Rather, itis a failure on the part of both the controllers and pilots to utilize the ATC system properly and to its maximum capability. ¢ ember Member 12/ "Q. .. . What is your definition of final approach course? "A, [FAA witness, Chief, ATC Operations Procedures Division] Well, I suppose it would vary, depending on where the aircraft was told to intercept the final approach course. It would extend from that point in towards the runway. "OQ. Could the final approach course be 85 miles long? "A, Conceivably. Surely." (Tr. 2379) APPENDIX A. Investigation and Hearing --- Footnotes: [12/ The FAA issued Advisory Circular 0046, "Aviation Safety Reporting Program! on May 9, 1975. The Advisory Circular states that the program will serve as a basis for an evaluation study of the National Air Transportation System by providing reporting procedures and by inviting pilots, controllers and other users of the airspace system or any other person to report discrepancies or deficiencies noted in the system to the FAA. The program will initially apply to that part of the system involving the safety of aircraft operations, including departure, en route, approach and landing operations and procedures; air traffic control procedures, pilot/controller com- munications; the aircraft movement area of the airport and near midair collisions.]

[ANALYSIS] Pilots should not be faced with the necessity of choosing from among several courses of action to comply with a clearance. . The Board believes that the clearance, under these circumstances, should have included an altitude restriction until the aircraft had reached a segment of the published approach procedure or the issuance of the approach clearance should have been deferred until the flight reached such segment. Therefore,the Safety Board concludes that the clearance was inadequate and its issuance and acceptance was the result of a misunderstanding between the pilot and the controller. $ The Board believes that there is a general lack of understanding between pilots and controllers in their interpretations of air traffic control-procedures. There is also a lack of understanding about the meaning of some words and phrases used by both the controller and pilot in the handling of IFR traffic in the terminal area. In this case, there was no definition of the term "radar arrival" or "final approach course," nor, as indicated earlier, did there seem to be common understanding between pilots and controllers as to the meaning of "radar control. "' Therefore, the Safety Board concludes that it is essential that a lexicon of air traffic control words and phrases be developed and made available to all controllers and pilots who operate within the National Airspace System. Additionally, there should be one book of procedures for use by both pilots and controllers so that each will understand what to expect of the other in all air traffic control situations. This manual must be used in the training of all pilots and controllers. The need for such a lexicon and procedures manual is evident from the circumstances of this accident. Flight 514 was vectored to intercept the 300° radial of Armel, the reciprocal course of which coincides with the course for the intermediate and final approach segments of the published instrument approach procedure. The vector was given when the flight was more than 80 miles from the airport and at a point where the 300° radial of Armel was not a part of the published instrument approach procedure. While proceeding inbound on the 300° radial ui Armel, the flight would not have reached a segment of the published approach procedure until it arrived at Round Hill. However, there was some testimony contending that Flight 514 was on its final approach course when the flight intercepted and was inbound on the 300° radial, and accordingly it was permissible for the pilot to descend to the minimum altitude of 1,800 feet prescribed for crossing the final approach fix of the VOR/DME instrument approach procedure. Qualified instrument pilots and air traffic controllers should know and understand beyond equivocation that the coincidence of the inbound course being an extension of the final instrument approach course does not permit ‘descent to altitudes lower than those published for that air space segment unless specifically authorized by ATC. A clear, precise definition of final approach course and final instrument approach course should preclude future misunderstandings. Neither of these terms was defined in the AIM at the time of this accident.

[ANALYSIS] Shortly after 1107, the captain first expressed doubt concerning the action he should be taking and the minimum altitude to which he was descending. He noted that the minimum altitude to Round Hill (from Front Royal) was 3,400 feet. He discussed the chart with the crew and again decided that the flight was authorized to descend to I 1,800 feet, the intermediate approach segment altitude. Seconds later the altitude alert system warning sounded indicating that the flight was approaching 1, 800 feet and the captain stated that he had seen the ground ''a minute ago.'' The first officer indicated that he had seen the ground also. Apparently they had only fleeting glimpses of the ground and did not derive any relative altitude information from what they saw. The first officer mentioned the power and the captain noted that they had a high sink rate. Then the captain said that the ground should be visible in just a minute. At 1108:57, the altitude alert sounded again. This sound may have been caused by a pilot positioning the altitude alert control to cancel further warnings. This is a normal TWA procedure once cleared to descend below the initial approach altitude. In this particular case the aircraft had arrived at the altitude the captain had determined to be the initial approach altitude, and clearance for the approach had been received. Subsequent altitude information was provided by the barometric altimeter and height-abovethe-ground information was provided by the radio altimeter. There was some conversation regarding a downdraft and the radio altimeter warning horn sounded then stopped. The captain said at 1109:20, ''Get some power on.''! The radio altimeter warning horn sounded again and at 1109:22, the sound of impact was recorded. The first radio altimeter warning was activated by the aircraft coming within 500 feet of the terrain, the designated altitude where the radio altimeter will begin to indicate the altitude. The second radio altimeter warning sounded as the aircraft approached 100 feet above the terrain. TWA's procedure, when conducting a nonprecision approach, requires that the radio altimeter be set to provide a warning at 100 feet above the terrain. The first warning came 7 seconds before impact and the second warning about 1 second before impact, after the captain ordered the first officer to '' get some power on.'' The crew should have realized that the aircraft should not have been that close to the ground at that point in the approach. However, their reaction to the warning probably could not have been faster than it was. A review of the flight data recorder graph indicates that at the times when the recorded altitude can be cross-checked against other altitude data sources within the aircraft, the aircraft was near “2.27 - the altitudes recorded. This indicates that the altimeter system was operating properly. The elevation at impact was about 1,675 feet. The altimeter was set at 29.70, the last altimeter setting given to the crew. Two reasons why the aircraft might have been below its target: altitude of 1,800 feet are evident.

[ANALYSIS] At the Safety Board's public hearing, FAA witnesses testified that they were not aware that there was any potential misunderstanding on the part of pilots as to the meaning of the term '"'cleared © for the approach, '' in a case where a nonprecision approach is made, . particularly when the clearance was issued a long distance from the airport. The evidence, however, does not support this conclusion, since, for several years prior to this accident, various organizations had perceived a problem in the use of the term ''cleared for the approach, "' Ironically, approximately 6 weeks before the TWA accident an air carrier flight, after being ''cleared for the approach, '' descended to 1,800 feet while outside of the Round Hill intersection during a VOR/DME approach to runway 12 at Dulles. The carrier involved had implemented an anonymous safety awareness program, was in fact made aware of the occurrence, and subsequently issued a notice to its flightcrews to preclude the recurrence of a near-fatal misinterpretation of an approach clearance. The Board is encouraged that such safety awareness programs have been initiated. It is through such conscientious safety management that the expected high level of safety in air carrier operations can be obtained. In retrospect, the Board finds it most unfortunate that an incident of this nature was not, at the time of its occurrence, subject to uninhibited reporting and subsequent investigation which might have resulted in broad and timely dissemination of the safety message issued by the carrier to its own flightcrews. Both the USAF and TWA had pointed out to the FAA that the terminology "cleared for the approach"! could be misinterpreted and ~ 30 - that pilots might, understand that they could descend unrestricted unless a specific altitude restriction was included in the clearance. With respect to the crew of TWA 514, the conversation in the cockpit as reflected inthe CVR transcript permits no other conclusions than that they assumed the clearance received permitted an unrestricted descent to 1,800 feet. Subquestions requiring discussion are whether other available information should have indicated to the crew the .unsafe nature of such a descent and why the crew was not alerted at least to the point of making inquiry to ATC. Considering the number of times the captain examined this chart after being informed that he was to divert to Dulles, he should have realized that the minimum altitude of 1, 800 feet might not be a safe altitude. Although the captain did not know his exact position - relative to the terrain when he received the approach clearance, the Board believes that with his VOR tuned to Armel and with the information provided by that navigational aid, he should have been able to read his DME range from Armel. At the time he received the clearance, he was about 44 nmi from Armel on the 300° radial inbound to the station. By reference to the approach chart, he should also have been able to identify the high obstacles between that position and the. Round Hill intersection.

[ANALYSIS] The pilot should question a clearance which leaves any doubt as to what - 36 -. 13/ course of action should be followed. —- The Board also believes that it is incumbent upon the controller to ascertain beyond a doubt that the terminology of a clearance conveys the intent to the pilot, and to question the pilot if there is any doubt that he has understood it and is initiating actions compatible with the intent of the clearance. Since, as FAA witnesses testified, the ATC system is a cooperative system, it is imperative that pilots and controllers fully understand the intent and execution of clearances to the extent that one is able to back up the other whenever there is doubt that the clearance or the execution of it may be unsafe or is likely to lead to an unsafe situation. 2.2 Conclusions a. .Findings 1. The flight operated without reported difficulty and in a routine manner until the diversion to Dulles Airport from Washington National Airport was approved. 2. The crew of Flight 514 reviewed the approach chart for the VOR/DME approach to,runway 12 at Dulles several times before beginning the approach. 3. The Washington Air Route Traffic Control Center controller vectored the flight to intercept the 300° radial of the Armel VOR at a point about 80 nmi from the VOR. This portion of the radial was not part of the published instrument approach. 4. The crew of Flight 514 intercepted the radial and tracked inbound on it, and control of the flight was passed to the Dulles approach controller. 13/ Subsequent to the accident the FAA amended 14 CFR 91.75(a) to reemphasize that "If a pilot is uncertain of the meaning of an ATC clearance, he shall immediately request clarification from ATC." 10. ll. 12, The Dulles approach controller cleared the flight for a VOR/DME approach to runway 12 when the aircraft was about 44 nmi from the airport. The clearance contained no altitude restrictions. The captain assumed that the flight could descend to 1,800 feet, immediately. The first officer, who was flying the aircraft, initiated an immediate descent to 1, 800 feet. The flight encountered icing and turbulence during the descent. Neither of these conditions should have appreciably endangered or restricted the con- ‘trol of the aircraft, but contributed in the apparent inability of the crew to arrest the descent at 1, 800 feet. The first officer allowed the aircraft to descend below the target altitude of 1,800 feet and did not take sufficient corrective action to regain and maintain that altitude. The first officer's altimeter was set properly. It is possible that wind velocity over the hilly terrain may have induced an altimeter error which could have caused the instrument to indicate that the aircraft was higher than its actual altitude. However, the crew's last.comments regarding altitude indicated that they knew they were below 1,800 feet. --- Footnotes: [13/ course of action should be followed. —- The Board also believes that it is incumbent upon the controller to ascertain beyond a doubt that the terminology of a clearance conveys the intent to the pilot,]

[ANALYSIS] As a result, the dark conditions would have significantly degraded the flight crew’s ability to see and safely avoid terrain while making the visual transition from the instrument approach to an intended landing. The reduced visibility and light snow showers near the airport would have further degraded the flight crew’s ability to see and safely avoid the terrain. This accident reveals that the aeronautical definition of night does not adequately take into account darkness in mountainous terrain. Specifically, because pilots do not have sufficient ambient lighting to see and safely avoid unlighted terrain during periods of darkness before official night, the night restriction on the VOR/DME-C approach is not enough to ensure safety of flight. Thus, nighttime restrictions may not sufficiently mitigate potential hazards associated with flight operations into ASE and other airports with mountainous terrain during periods of darkness.61 60 On February 21, 2002, an ASE tower controller observed the sun fully set behind the terrain about 1725. At that time, the elevation of the sun was 4.2º, resulting in a visible sunset that was about 26 minutes earlier than the official sunset of 1751. On the night of the accident, the sun was at an elevation of 4.2º about 1803, resulting in a visible sunset that was about 25 minutes earlier than official sunset. 61 To address this issue, the Safety Board issued Safety Recommendation A-02-08 on April 15, 2002. Safety Recommendation A-02-08 asked the FAA to “revise any restrictions and prohibitions that currently reference or address ‘night’ or ‘nighttime’ flight operations in mountainous terrain so that those restrictions and prohibitions account for the entire potential period of darkness or insufficient ambient light conditions, and establish a method to clearly communicate to flight crews when such restrictions and prohibitions apply.” NTSB/AAB-02/03 40 SUMMARY 1. The flight crew made numerous procedural errors and deviations during the final approach segment of the VOR/DME approach to Aspen (ASE). • The flight crew crossed step-down fixes below the minimum specified altitudes. • The flight crew descended below the minimum descent altitude (MDA), even though airplane maneuvers and comments on the cockpit voice recorder (CVR) indicated that neither pilot had established or maintained visual contact with the runway or its environment. • Contrary to the airplane manufacturer’s procedures, the captain deployed the spoilers after the landing gear had been extended and the final landing flaps had been selected, and he set engine power to 55 percent N2 rather than 64 percent N2. • When the airplane was 1.4 miles from the runway (about 21 seconds before the accident), the captain asked, “where’s it at?” but did not abandon the approach, even though he had not identified, or had lost visual contact with, the runway. • Radar data and CVR comments indicated that, until the airplane began turning to the left about 10 seconds before the accident, the flight crew probably did not have the runway or its environment in sight. 2. The crew demonstrated poor crew coordination during the accident flight. • The captain did not discuss the instrument approach procedure, the missed approach procedure, and other required elements during his approach briefing because he expected to execute a visual approach to the airport. • The captain and the first officer did not make required instrument approach callouts, and the first officer did not call out required course, fix, and altimeter information. • The flight crew did not discuss a missed approach after receiving a third report of a missed approach to the airport and a report of deteriorating visibility in the direction of the approach course. 3. The flight crew was under pressure to land at ASE. • Because of the flight’s delayed departure from Los Angeles International Airport and the landing curfew at ASE, the flight crew could attempt only one approach to the airport before having to divert to the alternate airport. • The charter customer had a strong desire to land at ASE, and his communications before and during the flight most likely heightened the pressure on the flight crew.

[ANALYSIS] NTSB/AAB-02/03 41 • The presence of a passenger on the jumpseat, especially if it were the charter customer, most likely further heightened the pressure on the flight crew to land at ASE. 4. Darkness, reduced visibility, and light snow showers near the airport at the time of the accident significantly degraded the flight crew’s ability to see and safely avoid terrain. 5. The March 27, 2001, Notice to Airmen (NOTAM) regarding the nighttime restriction on the VOR/DME-C approach was vaguely worded and ineffectively distributed. • The NOTAM stated, “circling NA [not authorized] at night,” but the intended meaning of the NOTAM was to prohibit the entire instrument approach procedure at night. • Pilots might have inferred that an approach without a circle-to-land maneuver to runway 15 was still authorized. • If the FAA had worded the first NOTAM more clearly, it might have made more of an impression on the first officer when he received the preflight briefing from the Automated Flight Service Station and might have affected the conduct of the flight. • The local controller could not notify the flight crew of the NOTAM because the Denver Center had not sent a copy to the ASE tower. PROBABLE CAUSE The National Transportation Safety Board determines that the probable cause of this accident was the flight crew’s operation of the airplane below the minimum descent altitude without an appropriate visual reference for the runway. Contributing to the cause of the accident were the Federal Aviation Administration’s (FAA) unclear wording of the March 27, 2001, Notice to Airmen regarding the nighttime restriction for the VOR/DME-C approach to the airport and the FAA’s failure to communicate this restriction to the Aspen tower; the inability of the flight crew to adequately see the mountainous terrain because of the darkness and the weather conditions; and the pressure on the captain to land from the charter customer and because of the airplane’s delayed departure and the airport’s nighttime landing restriction.

[ANALYSIS] Instead, the flight crewmembers might have viewed this information as a barrier to achieving their intended goal of landing at ASE, causing them not to recognize the increasing evidence supporting the need for an alternate course of action. Visual Factors When the VOR/DME-C instrument approach procedure was first established at ASE in December 1988, the procedure was not authorized at night. The FAA removed the night restriction from the procedure in October 1994 because of complaints from user groups. In March 2001, the instrument approach procedure was again restricted at night after an FAA flight inspection determined that areas of unlighted terrain conflicted with traffic patterns and circling descent maneuvers near the airport. The night restriction on the VOR/DME-C instrument approach procedure did not restrict the procedure during evening civil twilight. However, the decrease in ambient illumination during civil twilight results in a decrease in contrast between objects and the background scene, which diminishes a pilot’s ability to visually distinguish between terrain features and the sky and to visually detect unlighted objects. The amount of ambient light decreases by two orders of magnitude (that is, about 100 times) from the beginning to the end of evening civil twilight. Also, the sky during civil twilight can be much brighter than terrain features, and pilots may be exposed to higher ambient light levels at altitude before descent. Pilots may also experience rapid decreases in ambient illumination during approach and descent, especially in mountainous areas where terrain features may rise above the horizon and reduce the amount of ambient illumination at lower elevations. The mountainous terrain surrounding ASE reduces the amount of ambient lighting during evening civil twilight at lower elevations, including positions on and near the airfield. Although this accident occurred only 7 minutes after the end of evening civil twilight and the beginning of night, the mountainous terrain created twilight and nighttime conditions much earlier. The controller-in-charge stated, in a postaccident interview, that the sky was “very dark” in the minutes before the accident. In fact, the 59 C.D. Wickens, Engineering Psychology and Human Performance, 2nd ed. New York: Harper Collins, 1992. NTSB/AAB-02/03 39 Safety Board calculated that the sun had set below the mountainous terrain about 25 minutes before the official sunset time60 and that evening civil twilight ended about 1830 rather than 1855. Also, the shadow for the ridge immediately to the west of the accident site would have crossed the accident site 79 minutes earlier than official sunset. As a result, the dark conditions would have significantly degraded the flight crew’s ability to see and safely avoid terrain while making the visual transition from the instrument approach to an intended landing. The reduced visibility and light snow showers near the airport would have further degraded the flight crew’s ability to see and safely avoid the terrain. This accident reveals that the aeronautical definition of night does not adequately take into account darkness in mountainous terrain.

[ANALYSIS] Thus, the flight crew was not adequately prepared to perform ASE’s instrument and missed approach procedures. In addition, Avjet’s manual indicates that the captain was to brief the airplane’s configuration and approach speed, final approach fix altitude, minimum descent altitude (MDA), visual descent point, circling maneuver, runway information, and abnormal conditions. The CVR did not record the captain briefing any of this information. While the airplane was descending into the terminal area, the flight crew attempted to locate a specific highway because a visual approach to the airport follows parallel to this highway, but that effort was mostly unsuccessful. Afterward, the captain informed the approach controller, “I can almost see up the canyon from here but I don’t know the terrain well enough or I’d take the visual.” The controller acknowledged this information and provided the flight crew with instructions for intercepting the final approach course. According to Doppler weather radar and weather satellite information, the cloud tops were near 16,000 feet while the airplane was on approach to ASE. After descending through this altitude, the airplane was in and out of the clouds. 47 The AFSS specialist did not read the NOTAM verbatim and told the first officer that circling minimums were not authorized at night. The addition of the word “minimums” made the information conform somewhat closer to the intent of the NOTAM because the NOTAM modified the VOR/DME-C procedure and not the landing maneuver. 48 As stated in the “History of Flight” section, the sample flight plans provided by Gulfstream showed that the amount of fuel aboard the accident airplane would have exceeded the amount required by the Federal Aviation Regulations. NTSB/AAB-02/03 31 Execution of the Instrument Approach Procedure According to the intent of the NOTAM, the instrument approach procedure to ASE was not authorized after 1855. As a result, the flight crew should not have been attempting this procedure to the airport. The CVR did not record any discussion between the flight crewmembers about the NOTAM. However, the flight crew was likely not aware that official nighttime began 3 minutes before the 1858 landing restriction at ASE. Also, the flight crewmembers might have believed that the NOTAM, because of its unclear wording, did not apply to a flight that would not have to execute a circling maneuver to runway 15. If the FAA had worded the NOTAM more clearly (as in the revision issued the day after the accident), it might have made more of an impression on the first officer when he received the preflight briefing from the Hawthorne AFSS and might have affected the conduct of the flight. Finally, it is also possible that the first officer did not brief the captain about the NOTAM and, therefore, that the captain might not have been aware of the NOTAM. About 1856:06, the approach controller cleared the flight crew for the VOR/DME-C instrument approach procedure. The controller did not know about the NOTAM because the Denver Center had not sent a copy to the ASE tower.

[ANALYSIS] Finally, it is also possible that the first officer did not brief the captain about the NOTAM and, therefore, that the captain might not have been aware of the NOTAM. About 1856:06, the approach controller cleared the flight crew for the VOR/DME-C instrument approach procedure. The controller did not know about the NOTAM because the Denver Center had not sent a copy to the ASE tower. If the tower had received the NOTAM, the controller would have been required to notify the flight crew of the NOTAM either verbally or on the ATIS broadcast.49 About 1856:42, the controller informed all airplanes on frequency that the visibility north of the airport was 2 miles. The VOR/DME-C instrument approach procedure required the flight crew to maintain at least 14,000 feet until passing the Red Table VOR, the initial approach fix, and 12,700 feet until passing the intermediate step-down fix located 3 DME miles south of the Red Table VOR. (See figure 1.) The Airplane Performance Study for this accident indicated that N303GA crossed the Red Table VOR and the intermediate step-down fix at the altitudes specified by the approach procedure. (See figure 3.) The airplane crossed the Red Table VOR about 1857:49 at an airspeed of 160 knots and then crossed the step-down fix about 1858:40 at an airspeed of 150 knots. However, by this time, the flight crew could no longer comply with the local regulation that required the airplane to be on the ground by 1858. In addition, the captain was no longer in compliance with Avjet’s policy that required the pilot-in-command to ensure that the flight was conducted “in complete compliance” with local regulations. About 1858:03, the pilot of N527JA (a Canadair Challenger 600 airplane, which was preceding the accident airplane into ASE) reported that he did not have the airport in sight and that he would be going around. (Two previous reports of missed approaches had been transmitted over the ASE approach control frequency, one about 1844:43 and one about 1853:35.) The three reports of missed approaches and the deteriorating 49 FAA Order 7110.65, paragraph 4-7-12(b), states that, “on first contact or as soon as possible thereafter, and subsequently as changes occur, inform an aircraft…of destination airport conditions that you know of which might restrict an approach or landing. This information may be omitted if it is contained in the ATIS broadcast and the pilot states the appropriate ATIS code.” NTSB/AAB-02/03 32 visibility to the north should have alerted the flight crewmembers that they might also need to execute a missed approach because of the weather. However, the CVR did not record any discussion between the crewmembers at this point regarding a possible missed approach.

[ANALYSIS] Given the repori:ed weather conditions, the pilots probably saw the airport and che first officer began the circling maneuver for alignment with runway 19 near the expected position--about 1.5 mi northeast of the threshold for runway 23 and at an altitude near MDA. With the airport and the lighted runway in sight at the beginning of the maneuver, the first of*icer's attention primarily would have been directed toward the maintenance of those visual references. Additionally, since there was no company requirement that the nonflying pilot call out airspeeds, altitudes, or rates of descent for visual flight below MDA, bo’.n pilots probably were concentrating on those visual references. When the aircraft was turned toward the west and the descent below MDA was begun, it is. probable that all four factors--low visibility, wind shear combined with the aircraft's forward c.g. condition, visual illusions, and type of approach--combined to produce a complex, unstabilized, and illusory approach proftle. The aircraft's entry into a diminishing headwind shear condition would have caused the aircraft's nose to pitch down and would have caused the descent flightpath angle and rate of descent to increase. As these effects materialized, it is likely that nonhomogeneous fog ccnditions were encountered, Under these conditions, the pilcts' reactions to the illusion created by the reduced visual range cculd nave caused adcitional increases in the descent flightpath angle and rate of descent, or at least, could have made the pilots confortable with the increases induced by the wind shear, Additionally, the circling maneuver itself, which is an inherently unstable maneuver, probably disguised pitch control forces and other forces that might have . provided the pilot with kinesthetic cues about the aircraft's actual position and condition. Based ou all the evidence, the Safety Board concludes that the aircraft encountered at least mcederat : wind shear and entered nonhomogeneous fog during its descent below MDA, and that the wind shear induced increases in the aircrafc's descent flightpath angle and rate of descent which, when combined with the increased pitch control forces associated with a forward c.g. and the visual illusions created by the aircraft's entry into the nonhomogeneous fog, resulted in a descent into the trees, far short of the runwey threshold.

[FINDINGS] Fiight 977 was conducting a circling approach to runway 19 at Cape May County Airport; the designated first officer was flying the aircraft from the left seat. The pilots relied exclusively on visual references to conduct the circling approach to runway 19 and while descending below MDA. The wind shear probably induced a higher-than-desired rate of descent and descent flightpath angle during the aixcraft's descent below MDA. The pilots probably were influenced by visual illusicns created by fluctuating -risibility in the nonhomogeneous Suge The visual illusions probably induced the pilots to accept a higher-than-desired rate of descent and descent flightpath angle. the pilots probably lost all visual references as the descent progressed or became visually disoriented and did not initiate missed approach procedures in time to avoid impact with the trees because they were not monitoring altitude instruments. The company had ao altitude awareness procedures for visual flight below MDA or DH. The Federal Aviation Adminstration's surveillance of Atlantic City Airlines was inadequate in that weight and balance conputations were not monitored and a formal wind shear training program did not exist. The accident was survivable in the passenger cabin. Survivability in the cockpit was marginal because penetration by trees destroyed integrity of the left side of the cockpit. While the use of a shoulder harness probably would have lessened the severity of the captain's injuries, the availability of a shoulder harness to the first officer would not have prevented his fatal injuries. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the flightcrew'’s lack of altitude awareness during a circling approach which permitted the aircraft's flightpath to deviate below a safe approach profile. The aircraft's rate of descent ard descent flightpath angle increased as a result of wind shear encountered during the visual approach below minimum descent altitude. The flightcrew did not recognize these fligintpath deviations because they were relying on visual references which were degradeu by nonhomogeneous fog and on kinesthetic cues which were adversely affecte? by the aircraft's forward center of gravity resulting from the improper, - loaded aircraft. Contributing to the accident was the lack of company procedures requiring altitude-callouts during the visual portion of an instrument approach. BY THE NATIONAL TRANSPORTATION SAFETY BOARD /s/ KAY BAILEY Acting Chairman /s/ FRANCIS H. McADAMS Member /s/ PHILIP A. HOGUE Member /s/ JAMES B.

[ANALYSIS] It fe is unifkely that two experienced and qualified captains who had flown the approach many tines would have misinterpreted their flight instziuments. ; sh Therefore, the analysis of the circumstances indicates that the pilots yo did not seek information from the flight fastruments and, consequently, were vulnerable to the combined effects of a number cf factors. Moreover, the Safety Board believes that the combined effects probably caused the ; accident and that no single factor alone would have produced the same result. These factors are: (1) Low visibility, (2) wind shear combined vith the aircraft's forward c.g. condition, (3) visual illusions, and (4) the type of approach. tn Meth ae te -fidleareme rs eset ae te oe eR area on the night of the accident were produced by advection fog; that is, fcg produced by the movement of warm moist air over colder ground or water, This type of fog tends to deepen at moderate surface wind speeds (5 to 15 kns). At wind speeds greater than 15 kns, the fog tends to a develop into low stratus 0: stratocumujus clouds. Tbe nixing action produced by moderate winds creates 4 nonhomcgenesus fog condition wherein horizontal visibilities fluctuate rapidly. <Aiso, turbulence may develop which will wake instrument flying and aircraft control more difficult. -~ 17 - Acccrding to the passengers of Flight 977, turbulence existed throughout the flight from NAFEC, including the final minutes of the flight. Also, the station manager cbserved fluctuating surface visibilities because cf both horizontal and vertical movement of the fog. Surface winds were moderate with stronger winds aloft. Therefore, the Safety Board concludes that visibiiities in the approach area were 7 essentially as reported but were variable because of nonhomogeneous fog i] cond tions. Also, the horizon was not visible because it was obscu-ed ' by darkness ard fog. t De sagt EOE PETE MOE Wind Shear Combined with the Aircraft's Forward c.g. Condition-- Bi An analysis of the weather conditions that existed in the Atlantic City Ry area on the night of the accident shows that wind shear existed at 1 w Moa altitudes and that the wind shear was associated with a warm front that : moved rapidly northeastward through the area. The wind shear measurement Bf made vy the FAA at NAFEC clearly defined two distinct layers of shear. A Although these measurements were made about 30 mi northeast of Cape May County Airport and were made about 37 min before ~he accident, . the warm front sloped toward the northeast and passed across the Cape 4 May area at the surface about 2340. Therefore, the Safety Board concludes Pw that similar wind shear conditions existed in the Cape May area when the _ accident occurred. Moreover, the wind shear probably existed at lower = altitudes and the magnitudes of the shears probably exceeded those treasured at NAFEC.

[ANALYSIS] The aircraft was cectificated, equipped, and maintained in accordance with regulations and approved procedures. There was no evidence of a pre-impact failure or nalfunction of the airccaft's structure, powerplants, flight controls, or systems. Although the witness who heard the aircraft pass near his ground position described brief intermittent engine sounds, the passengers were not aware of any variations in engine sounds before the aircraft struck the trees. Therefore, the Safety Board concludes that the intermittent sounds were produced after the propel}}2rs struck the first trees. OP mae a TE CER eh ieee LR” Ihe en vee rreR Snr mere N Although damage to the altimetera precluded Functional tests, 4 the static pressure lines to both altineters were clear and it is unlikely a that both altimeters would have malfunctioned simultaneously, With 4 regard to the difference between the reported altimeter setting of 29.74 “ in and the barometric setting found in the pilot's altimeter, 29.62 in, 7 if the latter setting had existed in flight, the aircraft's actual altitude would have been about 120 ft higher chan the altitude indicated on the altimeter. Therefore, this differenze could not account for the aircraft's lLower-than-normal altitude. According to aircraft performance data, the aircraft's c.g. condition would not have sertously affected controllability es sut the aircraft's pitch axis because adequate rosevp elevator trim was available to provide zero elevator control forces for the range of configurations, power selections, and airspeeds that prebably existed during Flight 977's approach. However, as shown in the performance analysis, the forward c.g. condition would have altered the pitch control forces needed to maintaln a constanc descent flightpath angie under certain circumstances. From this standpoint and, since undex the conditions prevalent during the approach, kinesthetic cues from pitch forces would have been important to the pilot, the Safety Board concludes that the aircraft's reinforced longitudinal ctability, particularly the increased elevator control forces required to deviate from a trimmed airspeed, Pd. resulting from the forward c.g. condition probably was a factor in the Mb accident. ee a eal Nm Ne at Tae BEIM ty oy Bo j bis ve wea! Since the atreraft crashed about 4,000 ft short of the runway and since there was no evidence of a alfunction of the flight instruments or of flightcrew disability, the pilots elther misinterpreted their an fligiit instruments or did not seek information from the instruments. It fe is unifkely that two experienced and qualified captains who had flown the approach many tines would have misinterpreted their flight instziuments. ; sh Therefore, the analysis of the circumstances indicates that the pilots yo did not seek information from the flight fastruments and, consequently, were vulnerable to the combined effects of a number cf factors.

[ANALYSIS] Finally, since the crash site vas well within the 15° viewing angle of the VA3I, the Safety Board concludes that the pilots probably lost all visual references that would have provided altitude information shortly before the aircraft struck the trees. ~ 20 - This accident clearly demonstrates that adverse factors can, without warning, combine and quickly place a pilet in a siruation where his senses are unreliable and his control of the aireraft is in jeopardy. Under these circuustances, his only recourse is to rely on information from the flight instrurnents. Since factors such as optical accommodation, instrument interpretation, and pilot reaction time become critical at low altitudes, the better source of instrument information is from oral communication by the pilot who is not flying the aircraft. For this reason, we believe that Atlantic City Airlines’ lack of altitude awareness procedures for visual flight below MDA or DH must be considered contributory to this accident. Additionally, we believe that the captain, knowing that the approach was begun under decreasing visibility conditions, should have been prepared to immediately execute a missed approach when visual references were degraded or lost. The first officer's injuries were typical of those associated with forceful impact with solid nonyielding objects. The severe head and internal injuries suggest that these injuries were caused by the trees which penetrated the left side of the cockpit. Although shoulder harnesses were not provided in the cockpit, the availability of such restraining devices would not have prevented the first officer's injuries. However, the captain probably would have received lesser injuries, and perhaps could have avoided the head injurfes and the internal injuries had a shoulder harness been available and worn. The extent of the damage to the occupied seats, both in the cabin and in the cockpit, indicates that the forces involved in the ceceleration of the aircraft equalled or exceeded the limits to which these seats are designed. 7/ Ye ig estimated that the mean decelerative forces in this crash were in the range of 12 to 15 G's. The fatal injuries sustained by the passenger seated in the first row on the left side of the cabin are typical of bodily contact with a solid object. While it is possible that the bulkhead in front of this passenger was forced back far enough to make contact, there is evidence that this passenger did not have her seatbelt fastened and was thrown against the bulkheed, causing contre-coup and chest injuries. The fatal injuries received by the passenger seated in the second row on the right aisle side were probably associated with the collapse of his seat. It is possible that this passenger sustained chest injuries when kis seat collapsed downward, causing his chest to contact the lower edge of thea seat in front of hin. --- Footnotes: [7/ 14 CFR 23,561, Emergency Landing Conditions.]

[ANALYSIS] Therefore, the Safety Board concludes Pw that similar wind shear conditions existed in the Cape May area when the _ accident occurred. Moreover, the wind shear probably existed at lower = altitudes and the magnitudes of the shears probably exceeded those treasured at NAFEC. Teer ME tthe tet nek oo ONES OM rapt Be PPE AEM BH Mt ses SM. - 4 Aircraft performance calculations based on the conditions : hup: ttesized show that, in moderate wind shear of 5 kns per 100 ft, the ’ aire.aft would have terded to pitch nosedown to maintain its trimmed : airspeed, and its rate of descent and descent flightpath angle would have increased significantly if the rfilot took no corrective action. The aircraft's tendency to pitch nosedown would have been reinforced by the increased longitudinal stability associated with its forward c.g. condition. To prevent the descent flightpath angle from incrvasing Pt without increasing the noseup pitch trim, it would have been nece3sary for the pilot to apply and hold substantial amounts of back pressure on the control column which, assuming a constant thrust condition, would have caused the airspeed to decrease. Because of the aircraft's at forward c.g. location, the amount of back pressure needed to maintain \ a coastant descerr flignhtpath angle would have increased by 56 percent over that required for a DHC--6 with a midrange c.g. location, which is approximately the location calculated by the rlightcrew. Consequently, unless the pilot was aware of the need f.ur substantially increased pitch control forces, the associated hinesthetic cues could have led him to ! use less pull force than needed to mainta‘n a cons? ant descent flightpath ; angie, and the aircraft would still have pitched nosedown in response to 7 the wind shear, but at a lesser angle than for the zero control force situation. ' a Be ete ne meee ne. Visual {{llusions--Visual illusicns within nonhomogeneous visual fields are well known hazards associated with a pilot's reliance on visual references to conduct an approach and tanding in conditions of low visibility. 6/ A pilot will be influenced by these illusions when his visual range is shortened by a sudden reduction in visibility, such as that encountered when the aircraft enters nonhomogeneous fog. The shortened visual range creates the illusion that the aircraft is too high and 13 going higher. Unless this illusion is recognized and conscilously resisted, the pilot wiil decrease the aircraft's pitch attitude (and increase the descent flightpath angle) in an attemrt to make the visual range increase and appear normal again.

[ANALYSIS] The shortened visual range creates the illusion that the aircraft is too high and 13 going higher. Unless this illusion is recognized and conscilously resisted, the pilot wiil decrease the aircraft's pitch attitude (and increase the descent flightpath angle) in an attemrt to make the visual range increase and appear normal again. Additionally, if the visual range is shortened to the extent that visuai references are lost completely, the pilot may believe that the aircraft's pitch attitude has increased substantially and he may further reduce the pitch attitude in an attempt to reacquire the visual references, which will induce high rates of descent from which recovery, at low altitudes, may be difficult if not impossible. Type of Approach--lhe typical procedure used by Atlantic City Airlines’ pilots, including the captain of Flight 977, for making a circling approach to runway 19 at Cape May County Airport consisted of the following in a DHC-6: While inbound to the airport from the FAF (Sea Isle VOR), descend the aircrait to MDA and slow it to 100 kns; when the aircraft is at 100 kns, extend wing flaps to 10°, and when the aircraft is 1 to 1.5 mi from the airport and the airport is visible, begin the circling maneuver. The circling maneuver consisted of a right turn to a westerly heading followed by a left turn to a southerly heading for alignment with runway 19. The pilot would descend the aircraft below MDA during the latter portion of the manuever with the expectation of placing the aircraft on a normal 3° approach slope when aligned witn runway 19. Under normal circumstances, the aircraft would be so aligned about 1/2 mi from the <hreshold. Consequently, to achieve the desired approach Slope position when lining up with the runway, the pilot would have to descend the aircraft about 320 ft below MDA. This circling approach is complex since it requires that the airplane be banked, turned, and descended simultaneously to place it in the proper position in space from which a landing can be completed. Moreover, the maneuver involves variable flight control furces, particularly pitch control forces, which make trimming for zero pitch control forces difficult and probably impossible. 67 “Pilot Factors Considerations in See-To-Land," Technical Report APFDLATRA 76. ~52, The Bunker Ramo Corporation, May 1976. 0 begat tee Sk Fe Rae cotiomy” twp take aay ily t ‘ i j ; 4 x Yada 9 wa - POPES Eg mcg cag enna ae Based on the location of the wreckage, descriptions of a typical VOR instrument approach to runwey 23 with a circle to land on runway 19, and the need tu land ov runway 19 because of the lack cf runway lights on tne other runways, the Safety Board concludes that the aircraft was on a circling approach to runway 19 wren it crashed.

[ANALYSIS] The radar data indicated that, beginning a little more than 1 minute after the helicopter first appeared on radar after the accident NTSB Aircraft Accident Report 46 flew erratically, crossing site. Dispatch records sh dispatcher and asked if she could hear him terrain as high as 12,500 feet before descending rapidly near the crash owed that, shortly after the erratic flying began, the pilot radioed the . After the dispatcher responded in the affirmative, the truck a mountainside.” Postaccident wreckage examination indicated that the ’s tail rotor skid tube was bent upward (toward the tailboom) from its original position cause there was no evidence of any preimpact structural, engine, or system failures with th equire clear identification of the timing and location of the initial collision with te pilot stated, “…I s accident helicopter and exhibited scratches and abrasions in all directions. In addition, the outboard 9 to 10 inches of the two tail rotor blades were missing and were not recovered along the main wreckage path, indicating that the helicopter did strike something prior to the identified wreckage field. The pilot continued to key his microphone, and on the dispatch recording he could be heard breathing rapidly for about the next 39 seconds. Be e helicopter, it is likely that the helicopter’s initial collision with terrain resulted from either 1) pilot geographic disorientation (lack of awareness of position) and a controlled collision with terrain, because the pilot likely could not see the surrounding terrain in the dark night100 IMC conditions or 2) pilot spatial disorientation and an uncontrolled collision with terrain, because multiple risk factors for spatial disorientation (the pilot’s lack of a helicopter instrument rating and lack of helicopter instrument flying proficiency, maneuvering in dark night conditions, and turbulence) were present during the accident flight.101 The first of these scenarios, geographic disorientation, would likely result in a relatively stable flightpath leading up to the initial collision with terrain followed by an erratic flightpath. Spatial disorientation and loss of control, on the other hand, would likely result in a more erratic flightpath before the initial collision. Using the radar data to distinguish between these two possibilities would r rrain. While it appears that the pilot reported to the dispatcher that he hit the mountainside shortly after the erratic flying began, without knowing the exact point of impact, it is unclear whether the erratic flying led to the impact or if the erratic flying occurred because of the impact. As a result, it was not possible to evaluate the relative likelihood of these two possible causal explanations.

[ANALYSIS] Orasanu, L. Martin, and J. Davison, “Cognitive and Contextual Factors in Aviation Accidents,” in E. Salas and ciates, 2001), pp. 209-225.] NTSB Aircraft Accident Report 50 disorientation, which could lead to loss of control and/or a controlled flight into terrain accident. Had the pilot performed an interim risk assessment and considered the external circumstances or discussed them with the spotter,107 the NMSP dispatcher, or SAR ground personnel, he would have been more likely to recognize the potential hazards associated with an immediate takeoff and might have delayed his departure from the remote landing site until more favorable conditions prevailed. The NTSB concludes that the pilot exhibited poor decision-making when he chose to take off from a relatively secure landing site at night and attempt VFR flight in adverse weather conditions. 2.3 F re complex or perceptually demanding,108 experts on cold exposu actors Affecting the Pilot’s Decision-Making This section addresses factors that could have influenced the pilot’s decision-making both before accepting the mission and when he took off from the mountain to return to SAF, including fatigue, self-induced pressure, and situational stress. The NTSB considered whether environmental and/or physiological factors related to the cold temperature or the high altitude might have degraded the pilot’s decision-making in this case. The pilot spent about 50 minutes searching for the hiker and carrying her up a steep slope in very cold, windy conditions, in freezing precipitation, while dressed only in an unlined summer-weight flight suit and undergarments. Although cold stress can degrade cognitive performance, especially for tasks that a re at the U.S. Army Research Institute for Environmental Medicine indicated that, based on the pilot’s level of exertion and the terrain, the pilot’s metabolic rate would likely have been fairly high, offsetting the cold weather’s effects and minimizing the risk of hypothermia. After reviewing the circumstances of this accident for evidence of any physiological effects from the high altitude (the pilot was operating in the unpressurized helicopter for more than 2 hours before the accident), U.S. Army Research Institute for Environmental Medicine personnel indicated that the altitude would have affected the pilot very little as well. Because the pilot lived at an altitude of about 6,000 feet, he would likely have been sufficiently acclimatized to operate at higher altitudes.109 Since the time at altitude was relatively short (about 3 hours), there was likely little hypoxic effect on the pilot’s cognitive function. 107 The SAR commander spoke with the spotter while the pilot was retrieving the hiker and urged the spotter to remain in place and wait for ground teams to arrive if it was not safe to take off. However, it is not clear that the spotter shared this information with the pilot when he returned to the helicopter; the spotter did not recall the pilot raisi formance in Cold Environments,” in D.E. Lounsbury, R.F. ng the possibility of remaining on the mountain overnight. 108 R.G.

[ANALYSIS] Additionally, based on the elevation of the targeted search area (estimated to be 11,700 feet), the pilot should have anticipated that the helicopter would be operating near the 102 Las Vegas, New Mexico, is located about 40 miles east of SAF and 34 miles southeast of the accident loca . viewed before accepting the acci tion 103 The NTSB was unable to determine which specific weather reports the pilot re dent mission. NTSB Aircraft Accident Report 48 upper limit of its hovering and/or landing performance capabilities. (The helicopter’s dual-engine hover ceiling was 11,800 feet.)104 The NTSB concludes that, when the pilot made the decision to launch, the weather and lighting conditions, even at higher elevations, did not preclude the mission; however, after accepting a SAR mission involving flight at high altitudes over mountainous terrain, with darkness approaching and with a deteriorating weather forecast, the pilot should have taken steps to mitigate the potential risks involved, for example, by bringing cold-weather survival gear and ensuring that night vision goggles were on board and readily available for the mission. the landing site. It is possible that the pilot initially expected the hiker to walk up to the helicopter landing site, an ey arrived at the landing site and that, when they exited the helicopter after landing to pick up the hiker, they encountered strong wind and sleet. landed, the dispatcher again reported that the hiker “did not want to move.” The pilot sub Although the pilot may have considered some personal restrictions regarding maximum altitudes, terrain characteristics, and winds that would permit a safe landing in the search area, no official NMSP risk assessment policy existed and, therefore, there was no evidence that the pilot considered such restrictions. For additional information regarding risk assessments, see section 2.4.1. 2.2.2 Decision-Making During the Mission About 2010, when the pilot finally located the lost hiker, she was in a small clearing in a wooded area, with no suitable landing site nearby. The pilot maneuvered above the hiker and told the dispatcher to instruct the hiker to walk in the direction he was flying to reach d, as a result, the pilot likely believed that they would be able to depart the remote landing site relatively quickly after landing. However, although the hiker was ambulatory, she indicated to the dispatcher that she was cold and could not see well enough to move toward the helicopter’s landing site. NMSP dispatch records show that, about 2015 (about 4 minutes before sunset), the dispatcher asked if the pilot could land on top of the hill and send the spotter down to retrieve the hiker. The pilot, sounding exasperated, said, “That’s about the only thing we’re going to be able to do.” According to the spotter, during the pilot’s efforts to evaluate the nearest suitable landing site, the helicopter encountered strong winds and turbulence below 200 feet agl, it was getting dark, and low clouds were approaching from the west, all of which would have made the landing more hazardous.

[ANALYSIS] The pilot, sounding exasperated, said, “That’s about the only thing we’re going to be able to do.” According to the spotter, during the pilot’s efforts to evaluate the nearest suitable landing site, the helicopter encountered strong winds and turbulence below 200 feet agl, it was getting dark, and low clouds were approaching from the west, all of which would have made the landing more hazardous. Although the incoming weather and the increasing darkness meant that the operation was growing increasingly risky, the pilot made several passes over the landing site and, after determining that a safe landing could be accomplished, proceeded with the landing. About 2030 (11 minutes after sunset), the pilot landed the helicopter on a ridge about 0.5 mile uphill from the hiker and at an elevation of about 11,600 feet. The spotter reported that they encountered moderate turbulence when th When the spotter contacted the dispatcher by cellular telephone after the helicopter had sequently called the dispatcher to clarify the hiker’s intentions, and the dispatcher told him that she believed that the hiker expected them to help her to the helicopter. As a result, about 104 The accident helicopter’s dual- and single-engine service ceilings were 19,600 and 13,100 feet, respectively. NTSB Aircraft Accident Report 49 2033, the pilot (who was wearing an unlined summer-weight flight suit) told the dispatcher he was going to “walk down the hil 105 l a little bit.” He indicated that he expected the weather conditions to deteriorate and stated, “…if it does that, I’ve got to get the [expletive] out of here.” The pil . y reported “a heavy overcast” with heavy rain within 30 minutes of the accident. These conditions indicated a strong likelihood of reduced visibility and the potential for was surrounded by high, rugged terrain that or departure and, about 9 minu d shelter, and the pilot could have periodically used its engines to generate heat as needed throughout the night. However, because the remote landing site was less than 15 ot added, “I’m not going to spend a lot of time or we’re going to have two search and rescues.” There is no evidence that the pilot took the time to consider his options; rather, he promptly left the helicopter and walked down the heavily forested slope to find the hiker without stopping to get his flashlight out of his flight bag. These communications and actions suggest that the pilot was feeling increasing stress as a result of the deteriorating conditions and that he was fixated on the goal of retrieving the hiker and taking off again as quickly as possible The spotter stated that the strong wind continued to blow while the pilot was recovering the hiker. By the time the pilot and hiker returned to the helicopter (about 50 minutes after the pilot left the helicopter to retrieve the hiker and more than 1 hour after sunset), the sleet had turned to snow, and the clouds had lowered. Other witnesses who were camping at a lower elevation nearb structural icing. In addition, the remote landing site was no longer visible.

[ANALYSIS] It is likely that the pilot’s nature in this regard, combined with his concern for the well-being of the hiker, created significant self-induced pressure for him to ensure that the mission was successfully completed, despite increasingly difficult conditions. In addition, the accident pilot’s relatively brief tenure (about 6 months) in the chief pilot position may have left him vulnerable to management pressure to accept missions. Postaccident interviews with NMSP aviation section pilots and management personnel indicated that the DPS cabinet secretary’s questioning of pilots regarding launch decisions and his evident displeasure when the NMSP pilots did not accept a mission when others did (such as the New Mexico National Guard) sent the message that he wanted NMSP aviation section pilots to accept SAR missions, without adequate regard for the potential risks involved. priorities and influence are further discussed in section 2.4.1.) 2.3.3 Situational Stress The stress associated with the mission may h rmination to depart icated elevated concern after landing on the mountain, as evidenced by his statements, “It’s gonna start snowing up here and if it does that I’ve gotta get the [expletive] out of here,” “I’m gonna spend a lot of time or we’re going to have two search and rescues,” and, “Just tell her to start blowing her [expletive] whistle and I’ll try to find her, okay?” These statements and the pilot’s use of profanity (which was absent during earlier communications with the dispatcher) 118 The pilot became aware of the hiker’s situation during the mission. NTSB Aircraft Accident Report 54 suggest that the pilot was experiencing increased situational stress as a result of the perceived challenge posed by the deteriorating conditions. Cognitive effects of stress can include narrow king pressures that contributed to a reduction in the safety of flight operations conducted by the NMSP ating the weather conditions, the approaching darkness, and the potential for pilot fatigue, he may have elected to bring a second ait until morning to search for the hiker. At the very least, a ent process would likely have prompted the pilot to mitigate potenti ing of attention, response rigidity, longer reaction time to peripheral stimuli, and increased errors.119 It is possible, therefore, that the “tunnel vision” created by acute situational stress caused the pilot to fixate on the goal of taking off from the remote landing site as soon as possible and to disregard mounting evidence that it was not safe to take off after he returned with the hiker. 2.3.4 Summary of Factors Affecting the Pilot’s Decision-Ma The NTSB concludes that the pilot decided to take off from the remote landing site, despite mounting evidence indicating that the deteriorating weather made an immediate return to SAF inadvisable, because his fatigue, self-induced pressure to complete the mission, and situational stress distracted him from identifying and evaluating alternative courses of action. 2.4 Organizational Issues A number of organizational and management issues, including NMSP aviation section staffing, pilot flight and duty time and rest period limitations, and SMS programs and policies, were identified in this accident investigation.

[ANALYSIS] Using the radar data to distinguish between these two possibilities would r rrain. While it appears that the pilot reported to the dispatcher that he hit the mountainside shortly after the erratic flying began, without knowing the exact point of impact, it is unclear whether the erratic flying led to the impact or if the erratic flying occurred because of the impact. As a result, it was not possible to evaluate the relative likelihood of these two possible causal explanations. The remainder of this analysis discusses safety issues related to the following: the pilot’s decision-making, flight and duty times and rest periods, NMSP staffing, SMS programs and risk assessments, communications between the NMSP pilots and volunteer SAR organization personnel, instrument flying, and flight-following equipment. 2.2 Pilot Decision-Making 2.2.1 Decision to Launch on the Mission An NMSP dispatcher contacted the accident pilot about 1756 (almost 3 hours after he had completed his normal 8-hour work shift) regarding use of the NMSP helicopter to aid in SAR efforts to find a lost hiker in the mountains northeast of Santa Fe. According to NMSP dispatch 100 As previously noted, there was no moonlight at the time of the accident. 101 Spatial disorientation is the mistaken perception of an aircraft’s attitude relative to the earth. NTSB Aircraft Accident Report 47 recordings, the pilot was initially reluctant to launch on the accident mission because he believed it was too windy; the pilot stated that he would prefer to go up to look for the hiker in the morning. The pilot’s concern about the windy conditions likely stemmed from a flight he operated with the full-time helicopter pilot a few hours earlier when the two pilots encountered winds gusting to near 40 knots on the ground in Las Vegas, New Mexico,102 and adjusted their return flight to SAF. However, minutes after the request to fly the accident mission, the pilot called the dispatcher back and told her that he had “checked the wind”103 and could “probably go up and take a look” for the missing hiker. The full-time helicopter pilot said that the aviation section pilots normally obtained the TAF and METARs for SAF using the NOAA Aviation Weather Center website rather than checkin er elevations in which the SAR efforts took place predicted broken clouds at 14,000 feet, layered clouds to 22,000 feet, with widely scattere he pilot likely believed that he could fly to the search area (which was only 20 nm from SAF) and return to SAF quickly and safely before dark. (Several preflight comments indicate ith the English language and the remote, wooded, and unfamiliar area in which she was lost. The SAR mission extended into inet secretary, when he was a pilot in the NMSP aviation r was carried on board, and he routinely carried a survival g the area forecast for a local flight. TAFs and METARs are only valid within a 5-mile radius of the airport, but they provide detailed local weather information. It is possible, therefore, that the accident pilot reviewed the TAF and/or METAR for SAF before the accident mission.

[ANALYSIS] By the time the pilot and hiker returned to the helicopter (about 50 minutes after the pilot left the helicopter to retrieve the hiker and more than 1 hour after sunset), the sleet had turned to snow, and the clouds had lowered. Other witnesses who were camping at a lower elevation nearb structural icing. In addition, the remote landing site was no longer visible. Yet the pilot quickly prepared the helicopter f tes after the pilot returned to the helicopter with the hiker, the helicopter was airborne again. According to the spotter, the pilot seemed to indicate that he intended to depart through a narrow path, a “tunnel in the clouds.” An interim risk assessment performed at this point may have indicated to the pilot that a different course of action would be more prudent. Even rudimentary consideration of the adverse weather conditions should have indicated to the pilot that it was no longer safe to take off and attempt to return to SAF at that time. At that point, the only safe option was to wait inside the helicopter at the remote landing site, contact SAR personnel for information and assistance, and wait for the weather conditions to improve. Although the temperature was near freezing, the helicopter provided goo minutes flying time from SAF, the pilot was likely very tempted to attempt to fly back to SAF rather than wait inside the helicopter for an indefinite period of time. The fact that the helicopter was airborne within about 9 minutes of the pilot’s return indicates that the pilot was still fixated on departing as soon as possible, and he did not spend much time considering alternative courses of action.106 Taking off in a helicopter in dark (moonless) lighting conditions, with marginal visibility, strong wind, turbulence, low clouds, the potential for structural and/or engine icing conditions, and surrounded by high terrain poses an unacceptably high risk of spatial and/or geographic 105 appropri G.A. Klein, eds., Linking Expertise and Naturalistic Decision Making (Mahwah, New Jersey: Lawrence Erlbaum Asso Although the dispatcher suggested that the spotter retrieve the lost hiker, the pilot, who was slightly more ately clothed for the conditions, hiked to retrieve her. 106 In this regard, the pilot’s decision-making performance during the accident mission is reminiscent of a category of decision-making error that researchers have labeled “plan continuation error.” Plan continuation error has been defined as “failure to revise a flight plan despite emerging evidence that suggests it is no longer safe.” [J. Orasanu, L. Martin, and J. Davison, “Cognitive and Contextual Factors in Aviation Accidents,” in E. Salas and ciates, 2001), pp. 209-225.] NTSB Aircraft Accident Report 50 disorientation, which could lead to loss of control and/or a controlled flight into terrain accident.

[ANALYSIS] NTSB Aircraft Accident Report 53 2.3.2 Self-Induced Pressure Self-induced pressure might also have contributed to the pilot’s decision to accept the mission and then take off from the remote landing site. The pilot was described by his colleagues as having an exceptionally high degree of motivation for work-related tasks. According to some statements, the pilot may also have had a tendency to act before thinking things through. Specifically, the NMSP fixed-wing pilot told investigators that the accident pilot tended to “act right away before thinking things out.” Further, the part-time helicopter pilot told NTSB investigators that he thought the accident pilot lacked “temperance.” Further, the accident pilot was likely aware that the former chief pilot was relieved of his chief pilot duties by the DPS cabinet secretary after he decided not to accept a high-risk mission that would have involved less experienced pilots. The pilot’s decision to accept the accident mission, while not inappropriate, was consistent with his understanding of NMSP management’s priorities. (Management ave also played a role in the pilot’s dete from the mountain. The pilot’s communications with the dispatcher ind not The pilot’s wife and other aviation section pilots described the accident pilot as being “heroic” and indicated that it was in his nature to take personal risks to try to save others. Although the pilot did not know that the hiker lacked warm clothing and other survival equipment when he accepted the mission,118 he likely recognized that it would take ground SAR teams a long time to reach her remote location. Because of this concern and his awareness of the cold nighttime conditions on the mountain, the accident pilot was likely concerned that if he did not accept the mission, the hiker would not survive on the mountain overnight. Furthermore, successful past rescue outcomes for the pilot (for example, his recent commendations for the successful rescue of a man trapped in a flooded arroyo) may have reinforced his tendency toward risk-taking in the line of duty. It is likely that the pilot’s nature in this regard, combined with his concern for the well-being of the hiker, created significant self-induced pressure for him to ensure that the mission was successfully completed, despite increasingly difficult conditions. In addition, the accident pilot’s relatively brief tenure (about 6 months) in the chief pilot position may have left him vulnerable to management pressure to accept missions.

[ANALYSIS] The SAR mission extended into inet secretary, when he was a pilot in the NMSP aviation r was carried on board, and he routinely carried a survival g the area forecast for a local flight. TAFs and METARs are only valid within a 5-mile radius of the airport, but they provide detailed local weather information. It is possible, therefore, that the accident pilot reviewed the TAF and/or METAR for SAF before the accident mission. The TAF in effect for SAF at that time indicated that, between 1800 and 2200, temporary conditions would exist for the next 4 hours that included a cumulonimbus cloud ceiling at 6,000 feet agl (about 12,000 feet msl). (If such conditions extended an additional 15 miles to the northeast, there was a potential for thunderstorms and mountain obscuration in the search area during the mission.) The area forecast in effect for the high d light rain showers and isolated thunderstorms and gusting wind until 2100. A little more than 2 hours of daylight remained when the accident pilot was notified about the mission, and he might have thought it would be a short mission. The pilot knew that the hiker was in communication with the NMSP dispatcher via cellular telephone, and he might have believed that the hiker could quickly guide him to her location. This belief and the fact that the weather at SAF was warm and clear and the wind was calmer than earlier in the day likely contributed to the pilot’s decision to accept the flight. Further, the mountains in the search area were visible from SAF, and the pilot would have seen that they were clear of clouds at the time of departure. T d that the pilot predicted that the mission would be a quick “in and out” flight.) However, the hiker was unable to provide the pilot with much of the useful guidance that lost hikers can typically provide (such as describing her position relative to the sun, nearby landmarks, or terrain features), likely in part because of her limited proficiency w nighttime. According to the DPS cab section, a supply of cold-weather gea kit and extra blankets on such missions. Additionally, based on the elevation of the targeted search area (estimated to be 11,700 feet), the pilot should have anticipated that the helicopter would be operating near the 102 Las Vegas, New Mexico, is located about 40 miles east of SAF and 34 miles southeast of the accident loca . viewed before accepting the acci tion 103 The NTSB was unable to determine which specific weather reports the pilot re dent mission.

[ANALYSIS] Variation of the wind direction would result in a change in the location of updrafts and downdrafts in relation to the ridge. In addition, the presence of a horizontal axis vortex 36The unbroadcast ATIS wind direction, velocity and gust factor would have been similar to that of the earlier ATIS broadcast. 78 on the lee side of the ridge could have produced a localized updraft along the flightpath of the airplane. The Safety Board concludes that, although the variable wind conditions at the time of the accident may have caused localized updrafts and downdrafts in the area, the DFDR data indicates that there were no large-scale updrafts or downdrafts that would have affected the accident aircraft. During the approach to runway 15, to the point at which flight 1572 struck the trees, the airplane would have encountered moderate turbulence and localized updrafts and downdrafts due to the interaction between strong low altitude winds and rough terrain along the flightpath. Windshear due to strong gusty low altitude winds also occurred following the tree strike, as the airplane was on approach to the runway. An estimated mean wind profile indicated a decreasing headwind as the airplane descended to the runway. Although windshear was occurring as the airplane approached and passed over the ridge line, it was the gustiness of the low altitude winds, rather than a small-scale weather feature, that significantly affected airplane performance. Airspeed excursions amounted to only about 10 knots. Further, a descent rate of about 1,100 feet per minute was initiated by the flightcrew from about 1,840 feet msl and was maintained until tree contact. The linear nature of the pressure altitude trace indicates that the airplane’s flightpath was probably not significantly affected by updrafts, downdrafts, or windshear. Such an effect would be seen as a deviation from the near linear pressure altitude trace. Therefore, the Safety Board concludes that the decreasing headwind shear seen in the estimated mean wind profile data was not severe enough to cause the flightcrew to deviate below the MDA. Along the approach to runway 15, cloud bases were near 2,000 feet with multiple cloud layers above, and the tops of the clouds were above 15,000 feet. Flight visibility was near 0 miles in the clouds and 2 to 3 statute miles below the lowest cloud base. The first officer reported "there's the runway straight ahead " at 0055:57. The airplane was about 2.1 miles from the end of the runway at this time. Moderate rain probably occurred along the approach to the runway with more intense rain near the runway. Given the above described conditions, the Safety Board concludes that the weather at the time of the accident was not severe enough 79 to cause the aircraft to deviate below the MDA, and did not contribute to the accident. 2.8.1 LLWAS Equipment The northwest LLWAS sensor was physically out of alignment by 38 degrees and was corrected subsequent to the accident.

[ANALYSIS] The Safety Board concludes that quality control was inadequate within the FAA for accurately resolving the height of the trees on the ridge line. Therefore, the Safety Board believes that the FAA should examine and make more effective the coordinating efforts of the flight inspection program and the procedures development program, with emphasis placed on ensuring quality control during the development, amendment, and flight inspection process for instrument approaches. The Safety Board also concludes that there is great value in flying non-precision approaches with a constant rate or angle of descent until the airport environment can be visually acquired, if the avionics aboard the airplane can safely support such a procedure. Therefore, the Safety Board believes that the FAA should evaluate TERPS design criteria for nonprecision approaches to consider the incorporation of a constant rate or constant angle of descent to MDA in lieu of step-down criteria. 2.6.1 Precipitous Terrain The TERPS Handbook states that consideration should be given to induced altimeter errors and pilot control problems in precipitous terrain that may result when winds are 20 knots or more over such terrain. No changes to the instrument approach procedure for runway 15 at BDL were made to account for precipitous terrain. Precipitous terrain is not defined in the TERPS Handbook. However, the BDL runway 15 approach is used primarily when the winds are such that their speed and direction preclude the use of the primary runway 06/24. Such conditions are likely to result in wind velocities in excess of 20 knots over the ridge line, which occurred the night of the accident. Such winds adversely affect airplane altimetry, and although it does not appear to have been a factor in this accident, the Safety Board concludes that the FAA should have, but did not, consider the issue of precipitous terrain when developing and modifying the approach to runway 15. 71 The Safety Board believes that the FAA should incorporate precipitous terrain adjustments in the runway 15 approach. In addition, the Safety Board believes the FAA should include a more comprehensive set of guidelines concerning precipitous terrain adjustments in the TERPS (FAA Order 8260.3B) Handbook, clarifying the definition of precipitous terrain, and establishing defined criteria for addressing the potential effects of such terrain. FAA flight inspections of instrument approaches are not normally flown during adverse wind and turbulence conditions, such as those on the night of the accident, because the flight inspection pilots must fly under visual flight rules (VFR) to observe man-made obstacles and high terrain. Therefore, the flight inspectors may not be fully aware of how such adverse conditions affect the safety of a particular instrument approach. Because the Safety Board is concerned that non-precision approaches at airports other than BDL may be adversely affected by wind and turbulence associated with precipitous terrain, the Safety Board believes that the FAA should review and evaluate the appropriateness of the let-down altitudes for all non-precision approaches that have significant terrain features along the approach course between the initial approach fix and the runway.

[ANALYSIS] Because the Safety Board is concerned that non-precision approaches at airports other than BDL may be adversely affected by wind and turbulence associated with precipitous terrain, the Safety Board believes that the FAA should review and evaluate the appropriateness of the let-down altitudes for all non-precision approaches that have significant terrain features along the approach course between the initial approach fix and the runway. Airline safety departments and pilot labor organizations, such as the Allied Pilots Association and the Air Line Pilots Association, should be consulted as part of this review. In addition, the Safety Board believes that the FAA should solicit and record user comments about difficulties encountered in flying a particular approach to evaluate approach design more accurately. 2.6.2 Approach Plate Terrain Depictions The single 819-foot obstacle depicted on the final approach course of most BDL runway 15 VOR approach plates could lead flightcrews to believe that there was one discrete obstacle, and that it was the only dangerous point on the final approach (see Figure 1). However, the Safety Board concludes that the entire ridge line is an obstacle, and that it and similar terrain close to other airports should be fully depicted on the appropriate approach charts. As an example, see Figure 4, the BDL approach plate used by British Airways. The Safety Board continues to believe, as reflected in Safety Recommendation A-96-102, following the accident near Buga, Colombia, that the FAA should require that all approach and navigation charts graphically present terrain information. 72 (BRADLEY INTL) WINDSOR LOCKS VOR or TACAN 15 fenecs SSA 25m 48 Popaa I IRight onto BDL 149A to BOC Ee I [2830 te DODAYAEDL 114 & 2000 I 1830 I FAF to MAP (THR) 4.8nm 4693062 1, Final approach track 4608 loft of extended centrale, O00 from threshold. 2, Tranginon: AGL to MISTR vee Tra26M. 1Onm. Man alt, 3500, 3.1700 trem DAUUN eampate af a 3.5" angle of deacam, 4. 0ME indicates 02a at thraahald. Figure 4.--BDL approach plate used by British Airways. 73 2.7 Air Traffic Control Factors 2.7.1 ATC Altimeter Setting Distribution Procedures The approach controller is required to issue the QNH (above sea level) altimeter setting on initial contact with an arriving flight, in accordance with the Air Traffic Control Handbook, FAA Order 7110.65. AAL flight 1572 first contacted the approach controller at 0043:41. The controller should have issued the current altimeter setting of 29.38 inches Hg. at that time. The controller said that the omission was inadvertent.

[CONCLUSIONS > FINDINGS] At the time of the accident, the indicated altitude (height above airport elevation) that the airplane’s QFE altimeter was indicating was about 76 feet too high (based on the altimeter setting received at 0030), resulting in the airplane being 76 feet lower than indicated on the primary altimeters. 13. Although the flightcrew did not use the most current QNH setting they had available (29.40 inches of Hg.) in the standby altimeter, this error did not affect the accident sequence of events because the flightcrew had the correct, but outdated, QFE setting (29.23 inches Hg.) in the altimeters they were using when the accident occurred. 87 14. If the first officer had monitored the approach on instruments until reaching minimum descent altitude (MDA) and delayed his search for the airport until after reaching the MDA, he would have been better able to notice and immediately call the captain’s attention to the altitude deviation below the MDA. 15. The excellent crew resource management and flight skills that the flightcrew used, as reflected on the CVR recording following their encounter with the trees, were directly responsible for limiting the number of injured passengers to one individual. 16. FAA quality control was inadequate for accurately resolving the height of the trees on the ridge line. 17. There is great value in flying non-precision approaches with a constant rate or angle of descent until the airport environment can be visually acquired, if the avionics aboard the airplane can safely support such a procedure. 18. The FAA should have, but did not, consider the issue of precipitous terrain when developing and modifying the approach to runway 15. 19. The entire ridge line on the final approach course to runway 15 at BDL is an obstacle and it, and similar terrain close to other airports, should be fully depicted upon the appropriate approach charts. 20. Considering the fact that the pressure changes were described by the weather observer as "pressure falling rapidly," and especially in light of the controller’s failure to issue the current altimeter setting (29.38 inches Hg.) upon initial radio contact, and his 0044:34 entry of 29.34 inches Hg. in the ARTS system while the accident aircraft was on his frequency, it would have been prudent for the approach controller to have issued the 88 altimeter setting changes as the airplane neared the airport. 21. Closure of the tower was a good managerial decision because the safety of people in the tower was compromised by the adverse wind and rain. 22. The TRACON supervisor’s communications with the flight were appropriate and aided the flightcrew. He acted in a professional manner and should be commended for his willingness to assist the flight under the circumstances. 23.

[ANALYSIS] The Safety Board concludes that if the first officer had monitored the approach on instruments until reaching MDA and delayed his search for the airport until after reaching MDA, he would have been better able to notice and immediately call the captain’s attention to the altitude deviation below the MDA. 32Hydrostatic equilibrium is a balance between gravity and pressure gradient force. Strong winds flowing over high terrain can result in nonhydrostatic equilibrium. 67 Further, if the flightcrew had computed a visual descent point (VDP) for their approach to runway 15, as described in the AAL flight manual, the DME associated with a VDP would have provided the flightcrew with a specific point in space to leave the MDA for landing. There would be no reason for the first officer or the captain to be looking away from the instruments and out the windscreen for the airport until just before or at the VDP. Both of them could have been concentrating upon the level off at the MDA. However, in this case, no VDP was calculated, the first officer began looking for the airport prior to reaching the MDA in the critical stages of the descent to MDA, and he was not adequately monitoring the flight instruments to serve as an additional backup to the captain. If the captain had planned for a VDP, he could have set up a more shallow rate of descent commensurate with reaching the target altitude at or just prior to the VDP. The reduced rate of descent would have allowed the captain to better monitor his descent progress toward the MDA. Also, a reduced rate of descent would have enabled the airplane to capture the MDA with less of a roundout (altitude loss) when the autopilot altitude hold button was depressed. A reduced rate of descent would also have given the captain a greater opportunity to arrest the descent at the MDA if he had manually taken control of the airplane. The captain stated that he attempted to “level off” the airplane at MDA, using the altitude hold button of the autopilot. However, that feature of the autopilot was not engaged until after the first officer uttered his “you’re going below your....” statement. The captain never took manual control of the airplane to either arrest his descent at the MDA or to initiate a more positive and immediate recovery to the MDA once he flew below this altitude. In summary, regardless of the outdated altimeter setting that affected the indicated altitude that the flightcrew observed, they allowed the airplane to descend about 309 feet below the indicated MDA for the instrument approach. The captain initially did not recognize the descent below MDA, and he failed to react immediately when he was alerted to the altitude deviation by the first officer. The Safety Board concludes that the flightcrew’s failure to maintain the required MDA until the required visual references identified with the runway were in sight directly caused this accident. 68 2.5 Actions After Tree Strike Regardless of the fact that the flightcrew descended below the MDA, their actions after the initial tree strike were noteworthy.

[CONCLUSIONS > FINDINGS] The Safety Board therefore believes that the FAA should develop a uniform policy on shoe removal during evacuations, and require that all operators train their flight attendants to issue commands during an emergency evacuation consistent with that policy. 85 3. CONCLUSIONS 3.1 Findings 1. The flightcrew had the proper FAA airmen certifications and were qualified in accordance with applicable regulations and company requirements. They received the proper amount of crew rest before the accident flight and did not appear to be under unusual psychological pressure. 2. The airplane was properly fueled; passengers and cargo were loaded in accordance with AAL weight and balance requirements. 3. The flight was released in accordance with AAL dispatch procedures. 4. The weather at BDL was at or above the required minimums for landing, and included overcast clouds, visibility restricted by moderate rain, and strong, gusty wind conditions. 5. There was no evidence of a malfunction of the pitotstatic system, the autopilot system, the ground proximity warning system, the windshear warning system, or any flight control system that could have contributed to this accident. 6. With regard to the operation of flight 1572, AAL weather forecasts, as documented in the SIGMEC, were substantially correct. 7. AAL’s FAA-approved system of providing flightcrews with more focused forecasts, in the form of SIGMECs, is a valid method of weather dissemination. 8. The pressure decrease of about 1 millibar shown by the National Center for Atmospheric Research simulation for 86 the time and place of the accident was not underestimated by a factor of 2 to 3 because the flight data recorder’s altitude trace did not contain an altitude spike that would have resulted from a significant atmospheric pressure change. 9. With the exception of a current altimeter setting, the flightcrew had adequate information concerning the weather at BDL as they began their descent to the airport. 10. Because the flightcrew knew that the atmospheric pressure was falling rapidly, they should have requested a current altimeter setting from the BDL approach controller when one was not given, as required, upon initial radio contact. 11. If the flightcrew had received a current altimeter setting from the BDL approach controller when the flight first contacted the approach controller at 0043, it most likely would have resulted in the aircraft being 40 feet higher than it was when it struck the trees, and it might have given the aircraft enough clearance to miss the trees. 12. At the time of the accident, the indicated altitude (height above airport elevation) that the airplane’s QFE altimeter was indicating was about 76 feet too high (based on the altimeter setting received at 0030), resulting in the airplane being 76 feet lower than indicated on the primary altimeters. 13.

[ANALYSIS] They are (1) the error resulting from differences between the true mean temperature of the column of air below the airplane and the mean temperature assumed for the air column in the standard atmosphere; and (2) the error associated with departure of the distribution of 31AAL procedures require flightcrews to compare the differences between the primary and standby settings to the field elevation. If there is a discrepancy, they would be expected to read the ACARS message again for the correct settings, or, most effectively, to ask the approach controller for a current setting. 66 pressure from hydrostatic equilibrium.32 The effect of errors resulting from these two sources was found to be insignificant in this case. 2.4 The Descent Below MDA According to the AAL DC-9 Operating Manual the pilot not landing is responsible for calling out 1,000 feet above field level, 100 feet above minimum descent altitude (MDA), and MDA. The first officer, who was the pilot not landing, called out: “There’s a thousand feet” at 0055:06. A correlation of the CVR with the FDR revealed that this 1,000-foot callout was made at around 1,140 feet agl, based upon the flightcrew’s altimeter setting of 29.23 inches Hg. (QFE). Although this mandatory callout was made, the first officer did not follow additional company procedures by also calling out 100 feet above MDA (1,008 feet above the field elevation). However, 5 seconds later, at 0055:11, the first officer stated to the captain, “now nine hundred and eight is your uh...,” which indicated that he was aware of the close proximity of the MDA (908 feet agl) to the 1,000 above field level callout. At that time, the airplane was about 1,050 feet agl. The captain replied, “right.” The first officer stated that he then looked out the airplane windshield to locate the airport. When he looked back at the instrument panel, he saw that the airplane had descended below MDA. At 0055:25, the first officer said, “You’re going below your.…” At that time the airplane was about 350 feet above the ground and 5 seconds away from contact with the trees. Information from the DFDR indicates that a constant rate of descent of 1,100 feet per minute was maintained until the first officer uttered his “you’re going below your….” statement. The Safety Board concludes that if the first officer had monitored the approach on instruments until reaching MDA and delayed his search for the airport until after reaching MDA, he would have been better able to notice and immediately call the captain’s attention to the altitude deviation below the MDA. 32Hydrostatic equilibrium is a balance between gravity and pressure gradient force.

[ANALYSIS] However, as previously stated, the Board believes that these events occurred because the flight crew did not use the available cues on the airport surface during taxi and did not cross-check and confirm the airplane’s position on the runway before departure. Also, the flight crewmembers engaged in a nonpertinent conversation during a critical phase of flight (taxi operations), which caused them to lose positional awareness. In addition, before the airplane arrived at the dark runway, the first officer briefed the outage of the runway end identifier lights, he recounted that lights were out during his arrival at LEX early on the morning that preceded the accident flight, and the flight crew most likely read the NOTAM in the flight release paperwork that indicated that the runway 4/22 centerline lights were out of service. This accident is not the first one involving a wrong runway takeoff in which pilots did not use existing cues to identify the airplane’s location on the airport surface or cross-check and verify the airplane’s position before takeoff. For example, the final report on the October 31, 2000, Singapore Airlines flight 006 accident (see section 1.18.4) stated that the pilots did not verify the airplane’s position on the taxi route as the airplane was turning onto the wrong runway and concluded that the flight crew lost situational awareness and took off from the wrong runway despite numerous available cues that provided information about the airplane’s position on the airport. However, the final report for that accident cited a possible reason for the flight crew’s actions: the pilots did not adequately review the taxi route to ensure that they understood that the route to the correct departure runway required passing a parallel runway that was under construction and was open only for taxi operations. Numerous other wrong runway events have occurred. The ASRS database showed 114 reports of incidents from March 1988 to September 2005 in which flight crews of turbojet airplanes lined up on the wrong runway for takeoff. The ASRS reports indicated that the pilots involved in these events, pilots of other aircraft in the area at the time, or air traffic controllers detected the mistake either before or after takeoff. Also, on October 30, 2006, an Alaska Airlines 737 departed from the wrong runway at SEA. According to postincident interviews, the controller instructed the flight crew to taxi the airplane into position and hold on runway 34C, but the captain assumed that the airplane would be taking off from runway 34R. After the airplane departed uneventfully from runway 34R, the controller informed the flight crew that the airplane had departed from the wrong runway. Most recently, on April 18, 2007, a United Airlines Airbus A320 taxied onto runway 27, which was closed, instead of runway 30, the assigned departure runway, at MIA and began the takeoff roll. A NOTAM, which was included in the flight release paperwork, and the ATIS information broadcast, which the flight crew received, indicated Analysis National Transportation Safety Board A I R C R A F T Accident Report 76 that runway 27 was closed.

[ANALYSIS] A NOTAM, which was included in the flight release paperwork, and the ATIS information broadcast, which the flight crew received, indicated Analysis National Transportation Safety Board A I R C R A F T Accident Report 76 that runway 27 was closed. Also, the flight crew reported that the airport charts were out and available. As the airplane proceeded down runway 27, the airplane’s nose wheel light illuminated a truck flashing its lights. The captain and the first officer stated that they observed the truck at the same time. Simultaneously, the controller queried the flight crew to determine whether the airplane was on runway 30. The flight crew rejected the takeoff, and the airplane continued uneventfully to its destination. The Comair flight 5191 accident and other wrong runway takeoff events demonstrate that all pilots are vulnerable to this and other types of surface navigation errors. Even when navigation tasks are straightforward and simple, there is a potential for a catastrophic outcome resulting from human error if available cues are not observed and considered during taxi and the airplane’s position is not cross-checked at the intended runway. Systemwide interventions, such as improved airport standard markings and technologies to improve pilots’ positional awareness while navigating airport surfaces, can help protect against human error during airport surface operations by providing a redundant display of critical information. Such interventions are further discussed in section 2.4. In addition, the Safety Board is concerned about the breakdown in sterile cockpit discipline in the sequence of events leading to this accident. The pilots’ actions with regard to sterile cockpit procedures, specifically, their nonpertinent conversation while the captain was taxiing the airplane, were uncharacteristic with past reports. A first officer who flew with the captain 2 days before the accident described the captain’s sterile cockpit discipline as very good and stated that, when a newly trained first officer was occupying the cockpit jumpseat during one of the flight legs, the captain briefed him about sterile cockpit discipline. The first officer was also described as having good sterile cockpit discipline. On February 7, 2006, the Safety Board issued Safety Recommendation A-06-7, which was the result of the Board’s investigation of the Corporate Airlines flight 5966 accident.

[ANALYSIS] Even though discrepancies existed between the airport chart and the external cues available to the pilots because of an ongoing construction project at the airport,157 the chart depicted the paved taxiway and runway surfaces at the time of the accident. Another available resource within the cockpit was the instrumentation, including the heading bug, which had been set to 227º to correspond to the magnetic heading for runway 22. This heading information, which was clearly presented on both flight crewmembers’ MFDs and PFDs, would have provided the pilots with a real-time cue of their orientation relative to runway 22. 156  Although the first officer’s attention was focused inside the cockpit while the airplane was taxiing to the departure runway, he would have had multiple opportunities to look outside the cockpit and monitor the airplane’s progress along the taxi route. Such monitoring would have helped the first officer gauge and pace his activities with the available time left for taxi. 157  Taxiway A north of runway 26 had been closed (as indicated by a local NOTAM) and barricaded, and taxiway A5 had been redesignated as taxiway A. Neither of these changes was depicted on the airport chart. However, no evidence indicated that either pilot was confused by the discrepancies. In addition, the pilots of SkyWest flight 6819 and American Eagle flight 882 used the same chart to navigate to runway 22. Analysis National Transportation Safety Board A I R C R A F T Accident Report 63 In addition, the flight crewmembers could have communicated with the controller if they became unsure of their position at any time.158 However, the CVR did not record any statement by either flight crewmember about being unsure of the airplane’s position on the airport surface at any time during the taxi and takeoff roll. The CVR also did not record any indication that the crewmembers had attempted to confirm the airplane’s position on the runway before beginning the takeoff roll. The taxi routing and cues available to the accident pilots were identical to those that were available to the pilots of the two regional jets (SkyWest flight 6819 and American Eagle flight 882) that departed before Comair flight 5191. The SkyWest and American Eagle pilots had no difficulty identifying, and successfully navigating to, runway 22 using the available cues,159 even with the differences in taxiway signage and chart labeling.160 In addition, even though the airport configuration at the time of the accident had been in place for 1 week, the Safety Board did not identify any reports about surface navigation problems at LEX during that time.161 The Safety Board concludes that adequate cues existed on the airport surface and available resources were present in the cockpit to allow the flight crew to successfully navigate from the air carrier ramp to the runway 22 threshold. Preflight Activities and Actions During the Taxi 2.2.2.2  Because the availability of cues and aids for the pilots’ wayfinding task was not a factor in this accident, the Safety Board examined the crew’s actions during the preflight and taxi phases of the flight’s operation to identify possible reasons for the error.

[ANALYSIS] The first officer was preparing for an opportunity to upgrade to captain and was described as someone who would have made a good captain because of his adherence to standard operating procedures. 155  Even though the FAA does not consider the SEA and MIA incidents (discussed in section 1.18.4) to be runway incursions, the Safety Board does consider these incidents to be runway incursions. Analysis National Transportation Safety Board A I R C R A F T Accident Report 62 The Safety Board examined possible reasons why the flight crew stopped at the incorrect hold short line and attempted to take off from the incorrect runway, as discussed in sections 2.2.2.1 through 2.2.2.4. Available Cues and Aids for Wayfinding 2.2.2.1  Pilots integrate many sources of wayfinding information, either alone or in combination, to establish their position and navigate to their intended destination on the airport. Wayfinding information sources include geographic knowledge of standard airport surface features that are common to all airports (such as terminal buildings, ramp areas, taxiways, and runways), knowledge of standard conventions for marking and identifying airport surface features, airport charts, and taxi instructions from ATC. Because both pilots were experienced, they should have been knowledgeable about basic airport geographic features and signage and surface marking standards (the Safety Board did not note any signage or surface markings at LEX that were not in compliance with FAA regulations), and they should have been skilled at interpreting airport charts. In addition to their knowledge about the specific route (taxiway A to runway 22), multiple external cues and features were available to the pilots to support their navigation to runway 22, and no evidence indicated that their view from the windscreen was obstructed. Observations from a CRJ-100 airplane after the accident demonstrated that, during nighttime conditions, taxiway location signs were visible along the full length of the taxi route, a runway 26 holding position sign adjacent to the runway 26 hold short line was visible, and the runway 26 numbers were visible. Also, evidence indicated that, at the time of the accident, the signs identifying the critical features along the taxi route (that is, the runway 26 holding position sign, the taxiway A extension across runway 26, and the runway 22 holding position sign) were illuminated and would have been visible to both pilots.156 The flight crewmembers had resources available to them within the cockpit to support their navigation to runway 22, including the Jeppesen airport chart. Even though discrepancies existed between the airport chart and the external cues available to the pilots because of an ongoing construction project at the airport,157 the chart depicted the paved taxiway and runway surfaces at the time of the accident. Another available resource within the cockpit was the instrumentation, including the heading bug, which had been set to 227º to correspond to the magnetic heading for runway 22.

[ANALYSIS] Preflight Activities and Actions During the Taxi 2.2.2.2  Because the availability of cues and aids for the pilots’ wayfinding task was not a factor in this accident, the Safety Board examined the crew’s actions during the preflight and taxi phases of the flight’s operation to identify possible reasons for the error. The flight crew proceeded from the Comair operations center to the air carrier ramp area, where two Comair CRJ airplanes were located. The crew initially boarded the wrong airplane,162 even though the tail number of the airplane to be used for the flight 158  Such communications would have been consistent with good CRM (while at towered airports) and company policy. In addition, this behavior would have been consistent with reports of the pilots’ past performance. 159  The SkyWest first officer stated that he became “momentarily confused” about the airplane’s orientation when he looked up while crossing runway 26 but that he was able to reorient himself after identifying the holding position sign for runway 22. 160  The SkyWest and American Eagle pilots did not recall seeing the low-profile barriers on taxiway A north of runway 26. The pilots’ lack of recall in this area indicates that the barriers were not significant elements in the pilots’ wayfinding and that the taxiway A closure did not encumber their navigation to runway 22. The pilots’ lack of recall does not indicate that the barriers were inconspicuous or difficult to see when approaching or crossing runway 26. In fact, observations in a CRJ-100 airplane after the accident indicated that the barriers would have been clearly visible to the pilots during the taxi. The pilots’ failure to recall these barriers is not unusual because people can have difficulty remembering things in their environment with which they did not interact. 161  After runway 26 was reopened (November 2006), two events occurred in which air traffic controllers informed pilots that their airplanes were on that runway instead of runway 22 (see section 1.10.4). However, an interview with the pilot involved in the first event indicated that he had stopped for preflight activities, was aware of the airplane’s position, and did not intend to depart on runway 26. An interview with the captain of the flight involved in the second event indicated that he became temporarily confused about the location of runway 22 but that the airplane was never lined up to take off from runway 26. An interview with the first officer of this flight indicated that he and the captain were never confused about the correct runway for takeoff. 162  The Safety Board does not know which pilot was the first one to board the wrong airplane. Analysis National Transportation Safety Board A I R C R A F T Accident Report 64 was included in the flight release paperwork, and started its APU. Although these actions likely consumed a portion of the crew’s available time at the gate, CVR evidence and interviews with ground personnel indicated that neither pilot appeared to be rushed or hurried as he completed required tasks.

[ANALYSIS] For fatigue to be considered a factor, performance deficiencies need to be clearly discernible and consistent with the known effects of fatigue, and any evidence supporting alternative explanations for such deficiencies needs to be considered. In this accident, the surface navigation error and the attempted takeoff from the wrong runway were the most significant errors made by the flight crew. Other errors included the flight crew boarding the wrong airplane, the first officer asking during his takeoff briefing about the runway in use and offering up “two four,” the first officer repeating some items on the before starting engines checklist after having already completed it, and the first officer misidentifying the flight as “Comair one twenty one” when informing the controller that the flight was ready to depart. In addition, the CVR recorded the first officer using a yawning voice for two consecutive statements beginning about 0603:02. The CVR recording also indicated that the captain corrected some of the first officer’s errors.177 For example, the captain stated “its two two” when queried about the runway in use and pointed out that the before starting engines checklist had been completed. Such corrections are inconsistent with a fatigue impairment. 176  The wives of the captain and the first officer reported that their husbands drank coffee, and the CVR recorded the captain (about 0553:04) asking the flight attendant for a cola. 177  About 0549:49, the controller provided the flight crew with a modified clearance to ATL. The first officer told the controller that he “missed” the arrival route part of the clearance, but he acknowledged the rest of the clearance. The controller repeated the arrival information, which the first officer acknowledged. Because the arrival route information provided by the controller was different than that in the flight release paperwork, the first officer’s request for the controller to repeat the arrival information cannot be considered an error but rather a request for clarification, which is a normal event. Analysis National Transportation Safety Board A I R C R A F T Accident Report 71 This investigation was constrained by a lack of available information to establish the pilots’ sleep activities before the accident flight. For example, the Safety Board was unable to determine how much sleep the first officer obtained on the morning of August 26, 2006, whether either pilot received supplemental sleep (in the form of naps) during the afternoon of August 26, or whether the pilots’ sleep time was uninterrupted.178 Also, the Board was unable to obtain reliable information about the nominal sleep needs for the pilots or their normal sleep and wake patterns. It is important to note, however, that both crewmembers seemed to be aware of the need to obtain adequate rest for the next day.179 Specifically, each pilot called his wife on the evening of August 26, 2006. The telephone call from the captain to his wife ended about 2127. The captain told his wife that he was planning on going to sleep afterward. The telephone call from the first officer to his wife ended about 2102. The first officer told his wife that he was going to bed early.

[ANALYSIS] This analysis discusses the taxi and attempted takeoff sequence and the associated human factor issues, survival factors, efforts to mitigate surface navigation errors, ATC staffing, and other issues related to the accident flight. Taxi and Attempted Takeoff Sequence 2.2  The pilots attempted to take off from a different runway than the airplane had been cleared to use. Section 2.2.1 details the events leading to the attempted takeoff, section 2.2.2 considers the pilot human factors that might have played a role in this surface navigation error, and section 2.2.3 explains the ATC human factors that might have played a role. Wrong Runway Departure 2.2.1  Before Taxi Activities 2.2.1.1  The Comair Operations Manual indicated that the captain was to conduct the taxi briefing. The briefing was to incorporate Comair standard taxi information, including that both flight crewmembers should have the appropriate airport diagrams out and available and that traversing runways required extra diligence. Comair standard taxi information was to be briefed in its entirety for the first flight as a crew and could be abbreviated to “Comair standard” for subsequent flights. The accident flight was the pilots’ first flight Analysis National Transportation Safety Board A I R C R A F T Accident Report 58 as a crew, but, about 0556:14, the captain stated “Comair standard” instead of all of the information in that portion of the taxi briefing, including that runway 26 was to be crossed while navigating to runway 22. The abbreviated briefing was contrary to company policy, and the Safety Board notes that it is prudent for pilots to fully conduct taxi briefings according to standard operating procedures. However, despite this abbreviated briefing (which occurred about 10 minutes before the accident), multiple and more salient cues existed to aid the flight crew while navigating to the runway, and the airport navigation task was relatively simple,146 as discussed in this analysis. The first officer was the flying pilot for the accident flight. About 0557:23, he briefed the taxi route as part of the takeoff briefing,147 stating, “let’s take it out and … take … [taxiway] Alpha. Two two’s a short taxi.” Although the CVR did not record either pilot explicitly referencing the airport chart, this statement is consistent with the first officer examining the chart because no specific taxi instructions had been provided to the flight crew. Also, the number of times that each crewmember had previously arrived at or departed from LEX was likely not sufficient to allow either one to have memorized taxiway identifiers and routes.

[ANALYSIS] Analysis National Transportation Safety Board A I R C R A F T Accident Report 67 At the time that the first officer began to increase thrust for takeoff, FDR data showed that the magnetic heading of the airplane was about 266º, which corresponded to the magnetic heading for runway 26. Figures 6 and 6a show the approximate configuration of the captain’s MFD and PFD, including the heading bug setting and magnetic heading indication, when the airplane was lined up on the runway 26 centerline. As shown in the figures, the heading bug was offset by 40º, providing a salient cue that the airplane was not lined up on the correct departure runway. The CVR did not record any awareness by the flight crewmembers about this offset. Multifunction Display Figure 6.  Source: Rockwell Collins. The display was configured as requested by the Safety Board. Analysis National Transportation Safety Board A I R C R A F T Accident Report 68 Primary Flight Display Figure 6a. Source: Rockwell Collins. The display was configured as requested by the Safety Board. The wayfinding task includes an ongoing cross-check between an airplane’s expected and actual position using available cues in the environment and aids in the cockpit. The CVR did not record any discussion by the flight crew about the need to cross-check the airplane’s position on the runway.171 The Safety Board concludes that the flight crewmembers failed to recognize that they were initiating a takeoff on the wrong runway because they did not cross-check and confirm the airplane’s position on the runway before takeoff and they were likely influenced by confirmation bias. 171  On March 23, 2007, Comair revised its operations manual to include a departure runway checklist item. Analysis National Transportation Safety Board A I R C R A F T Accident Report 69 Fatigue 2.2.2.4  The Safety Board examined whether pilot fatigue could have been a factor in this accident by assessing the preconditions that could allow for the development of a fatigued state and examining the nature of the pilots’ demonstrated performance deficiencies. Potential conditions that can lead to the development of a fatigued state include chronic sleep restriction, acute sleep loss, circadian disruption (work during times when one would normally be asleep), and time since awakening. The captain and the first officer received more than the minimum required rest periods during their respective trips in the days before the accident, and their flight and duty times in the week and month before the accident would not have precluded them from obtaining adequate sleep.

[ANALYSIS] For example, in its report on the August 16, 1987, Northwest Airlines flight 255 accident, the Board found that checklists (except for the before start checklist) were not being called for or accomplished according to company procedures and that, after pushback, the flight crew initiated conversations that were not related to duty requirements and diverted the crew’s attention from task-related activities. Regarding the role of cockpit discipline in these errors, the report stated, “it is the captain’s responsibility to structure the manner in which his crew will accomplish its duties … he must set the tone for how this information will be proffered.” The report further stated that a well-defined role structure in the cockpit reduces ambiguity about each crewmember’s responsibilities and when those responsibilities will be performed.183 In addition, data from the LOSA Collaborative184 showed that flight crewmembers who intentionally deviated from standard operating procedures were three times more likely to commit other types of errors, mismanage more errors, and find themselves in more undesired aircraft situations compared with those flight crewmembers who did not intentionally deviate from procedures. The Safety Board concludes that the flight crew’s noncompliance with standard operating procedures, including the captain’s abbreviated taxi briefing and both pilots’ nonpertinent conversation, most likely created an atmosphere in the cockpit that enabled the crew’s errors. Summary of Pilot Human Factors 2.2.2.6  Even with the discrepancies that existed between airport charts and signage, the navigational task that the pilots of Comair flight 5191 faced—taxi to runway 22 via taxiway A—was straightforward and inherently simple. The primary cues for this navigational task—the airport markings and signage at hold short positions—were accurate and available to the flight crew. Flight crews are responsible for maintaining positive control of an airplane during all phases of flight, including surface operations. An essential component to positive control is knowing the position of the airplane at all times. It was the accident pilots’ responsibility, once they were cleared to taxi, to safely maneuver the airplane to runway 22. Each pilot had experience and expertise to allow the successful completion of this task using only the standard airport marking conventions; that is, once the pilots detected and recognized signs off the air carrier ramp that identified taxiway A, they 183  National Transportation Safety Board, Northwest Airlines, Inc., McDonnell Douglas DC-9-82, N312RC, Detroit Metropolitan Wayne County Airport, Romulus, Michigan, August 16, 1987, Aircraft Accident Report NTSB/AAR-88/05 (Washington, DC: NTSB, 1988). 184  According to its Web site, the LOSA Collaborative is a network of researchers, safety professionals, pilots, and airline representatives collaborating to provide, among other things, oversight and implementation of LOSA and a forum for information exchange regarding LOSA. The referenced data were provided to the Safety Board in January 2007.

[ANALYSIS] However, 221  K. Cardosi and A. Yost, Controller and Pilot Error in Airport Operations: A Review of Previous Research and Analysis of Safety Data, Technical Report No. DOT/FAA/AR-00/51 (Washington, DC: Department of Transportation, 2001). Analysis National Transportation Safety Board A I R C R A F T Accident Report 98 after the accident, the controller acknowledged that it might have been possible for him to detect that the accident airplane was on the wrong runway if he had been looking out the tower cab windows. Administrative tasks should not have priority over radar and tower monitoring tasks. The Safety Board concludes that, if controllers were to focus on monitoring tasks instead of administrative tasks when aircraft are in the controller’s area of operations, the additional monitoring would increase the probability of detecting flight crew errors. Therefore, the Safety Board believes that the FAA should revise FAA Order 7110.65, “Air Traffic Control,” to indicate that controllers should refrain from performing administrative tasks, such as the traffic count, when moving aircraft are in the controller’s area of responsibility. Air Traffic Control Staffing 2.5  In August 2005, an operational error occurred during the midnight shift at RDU, which resulted in a loss of separation between a departing and an arriving airplane. At the time, one controller was performing radar and tower functions in the tower cab. The FAA attributed this incident to “controller situational awareness and actions,” specifically, the controller’s “failure to maintain constant surveillance and awareness of [radar] displays and traffic situation, failure to comprehend and project the future status of displayed data, and failure to provide attention to detail necessary for the safe operation of aircraft under control.” The FAA’s final report on the operational error described the traffic situation as “moderately difficult” because the controller was handling nine airplanes (one departure, six arrival, and two taxiing airplanes) and thunderstorms were in the airspace, which led the controller to vector one of the airplanes off of a standard arrival route. As a result of this incident, the FAA issued verbal guidance in late August 2005 indicating that all facilities with radar and tower responsibilities were to be staffed with two controllers during the midnight shift so that the functions could be split.222 In November 2005, the LEX air traffic manager notified his staff by e-mail about this guidance. Staffing during the midnight shift at LEX (0000 to 0800) was frequently not in compliance with this guidance. The midnight shift at LEX was staffed with one controller until January 2006. Between January and March 2006, the midnight shift was staffed with two controllers 40 percent of the time. In April 2006, the shift was again staffed with one controller because traffic had increased during the day and swing shifts and the second 222  As stated in section 1.10.2.1, in November 2006 the FAA issued Notice N JO 7210.639 to formalize this verbal guidance.

[ANALYSIS] With poor visibility you may begin to go on instruments and then sensory illusions can make you delieve your instruments are wrong. There is just one way to beat false interpretation of motion. Put your faith in your instruments and not in your seuses. Know what kind of teicks your senses can play on you, keep caim, and have confidence in your instrument panel. 2. ANALYSIS 2.1 General The airplane was properly certificated and had been maintained In accordance with applicable regulations and procedures, The pilot was properly trained, certificated, and qualified to conduct the filght. He had the proper medical certification, adequate rest, and there was no evidence of preexisting or incapacitating disease or pathology which would have affected his ability to conduct the flight. Examination of the engines and propellers revealed that the engines were developing high power at impact. There was no evidence that power loss or malfunction were causal factors in this accident. There was no evidence found to indicate preimpact failure or malfunction of the airplane structure, syetems, or flight controls; however, severe Impact and postimpact fire damage destroyed many components which therefore could not be examined. There was no indication that the pilot encountered difficultles during the approach until the airplane passed the immediate vicinity of the VOR station (missed approach point). Therefore, there is no reason to suspect that airplane or ground navigational equipment malfunctions were causal factors, nor is there reason to suspect pitot-static malfunctions were factors. The fact that the eyewitnesses observed the airplane fly over the airport at or near the proper course and altitude further discounts such possibilitics. The witness observations and the fact that the landing geur and flaps were up suggest that the pilot was executing a missed approach procedure when the airplane crashed. Because the company procedure for a nonprecision approach specified gear extended, flaps at the approach setting (159, and airspeed 120 KEAS, the Safety Board concludes from the flap and gear positions that the pilot was unable to land from the approach and that he raised the gear and flaps at or near the VOR and Initiated the missed approach procedure. From the location of the ‘ed light deseribed by the surviving passenger, the Safety Board concludes that she saw the landing gear in-transit Nght which on when the gear doors sre open during landing gear retraction. The Safety Board concludes that the weather conditions at the time of the accident probably wero below the landing minimums specified for the approach being conducted. Byewitness accounts and official observations indicated deteriorating celling wh Oy a A : feo og 7 oe edge 4 el WT aig PERE PEE CME POR Tee ee a Sees ite dee ey = eee Re ee sb mean ola aceite -[4- and visibility about the time of the accident. The cloud ceiling decreased from 490 fect obscured at 2000 to 200 feet obscured at 2015, Also, the visibility decreased from 1 mile to 1/2 mile in moderate snow and fog during the same period.

[ANALYSIS] The cloud ceiling decreased from 490 fect obscured at 2000 to 200 feet obscured at 2015, Also, the visibility decreased from 1 mile to 1/2 mile in moderate snow and fog during the same period. If the pilot had flown a precise ground track and altitude (MDA) to the missed approach potnat, and ff the ceiling was 400 feet and the visibility was 1 mile, the pilot should have acquired the necessary visual cues to complete or attempt a normal landing. The fact that he initiated a missed approach when over the virport suggests strongly that the celling and/or visibility was less than minimums when the airplane approached the runway, and the pilot was not able to obtain adequate visual refxrences to continue the approach to a successful landing. The pilot most ‘ixely did obtain a glimpse of airport features as the airplane passed near the VOR, adjacert to the terminal area, as evidenced by the one passenger's exclamation (seream) that there was an airplane on the runway. Since there was no airplane on the runway at inat time, the comment by that passenger most probably referred to an observation of the terminal grea to the left of the runway where Frontier Flight 815 and other aircraft were parked under bright lights. That probably was the only occasion the pilot may have had to observe airport features because the surrounding terrain was in darkness. He wither did not see the approach Nghts or runway lights in suffictent time to descead fron the MDA and land, or he lost sight of the runway once he left MDA and had to stop the cescent. In either case, the airplane was not In position to land and the airport was overfic wn in preparation for a missed approach procedure. The ground witness <bservations of the flightpath of Flight 104 vary somewhat regarding its height and ground track as the airplane passed the terminal. Height estimates varied from 100 feet to 400 feat above the airport, and the alrplane's location varied from over the VOR to almost over the terminal. The weather conditions at the time must be considered when ¢valuating the witness observations. As indicated by the post-accident weather observation, the vertical visibility was probably less than 400 feet when the airplane passed. Further, some witnesses stated that they lost sight of the airplane after it flew away from them, and before it struck the ground. Since the afrplane struck the ground about 3,600 feet from the witnesses, this would indicate that horizontal visibility was less than 1 mile. None of the witnesses observed the landing gear or other features of the airplane distinesly. Some only saw lights but others definitely saw the airplane fuselage and wings.

[ANALYSIS] Further, some witnesses stated that they lost sight of the airplane after it flew away from them, and before it struck the ground. Since the afrplane struck the ground about 3,600 feet from the witnesses, this would indicate that horizontal visibility was less than 1 mile. None of the witnesses observed the landing gear or other features of the airplane distinesly. Some only saw lights but others definitely saw the airplane fuselage and wings. Based on the assessment of horizontal and vertical visibility at the airport, on the fact that the pilot began the missed approach procedure, and on the observation of the witnesses that they did not see the afrolane for more than a glimpse, the airplane must have been quite low. Also, based 01 a consensus of the witness observations, and on the weather eonditions, it appears ttiat Flight 104 probably passed over the runway, in the vicinity of the VOR, between 200 and 300 feet. The probability that the airplane was below the MDA for the approach wher observed by the witnesses could be accounted for by the possibility that the pilot saw the runway lights, descended below the MDA for the landing, and then elther lost sight of the runway of realized that he could no: land, because he was not aligned or was too far down the runway. Regardless of the reason for the low altitude when passing the terminal, the pilot should have been able to complete a successful missed approach and the airplane should have stained considerable altliude before It arrived at the point where It crashed. In order to analyze tha possible explanations for the loss of altitude, the Safety Board calculated the rate of descent required to descend from the altitudes observed by the witnesses, when the airplane passed over the alrport to the point of initlal ground contact. The straight-lire distance from the VOR (missed approach point) to initkal ground contact was sbout 3,250 feet. The afr:peed assumed for the calculations was 120 KIAS, which was about 150 miles par hour true airspeed, or 220 feet per second (fps). Using these distances and speeds, the elapsed time from abeam the VOR to ground contact would have been about 14.5 seconds. Since she airplane turned right from about 030° to about 060° during the missed approach, and essuming a standard rate turn (3° per second), it would have required 10 seconds to compiete the turn. Therefore, tha actual ground track would have been en are and the actu) distance from the VOR to ground contact would have been greater than 3,250 feet. Consequently, the elapsed time from Yor passage to impact would have been slightly over 15 seconds for the assumed 120 KIAS, Using the conservative time of 15 seconds, the rate of descent from 300 feet above the alrport to impact would have been 1,20) feet per minute (fpm), or 20 fps; from 200 feet it would have been 800 fpm or 13.3 fps; and from 100 feet it would have been 400 fpm, or 6.6 fos.

[ANALYSIS] The attopllot could have been disconnected when the pilot pushed the disconnect button, or when he overrode it by pulling back on the control yoke. {{ the autopilot malfunctioned and initlated the descent, or the pilot made an improper input once he recognized the deviation, he would have disconnected the autopilot eitiner by overciding it or pushing the disconnect. Again, the evidence is inconclusive and therefore, the role of the autopilot in the events of this aceident cannot be reasonably assessed. Airframe _icing.--The Safety Board also cons.cered the possibility that airframe ice was the reason for the unwanted descent. The meteorological conditions were conducive to ice cccumulations and the pilots of Frontier Plight 815 reported that they encountered feing curing thelr approach a few minutes before Sun West Flight 104's approach. However, N41070 was certificated and equipped for flight into known feing conditions and the existing conditions were not so severe that. they should have caused serious problems. The pitot of Flight 104 should have been able to cope easily with the existing conditions. Therefore, although it Is possible that the pilot failed to remove ice buildups during the approach, and the resultant adverse aerodynamic effects resulted in the unwanted descent, the Safety Boar believes such factors ave not likely in this case. Intentional maneuver.~-Another possibility considered by the Safety Board, which could account for t+: ocoservations of the survivor and the ground observers, involves the possibility tha. the pilot intentionally pushed the nose over to "duck under” and remain visval in an attempt to fly a circling approach to runway 20. A cireling approach would involve a turn eway from the runway followed by a reverse turn beck to the runway. Sinea tne winds were calm, the pilot could have landed on the runway in either direction. The possibility that the pilot saw the terminal area as he passed by suppo:ts speculation that the pilot may have contemplated or even attempted such a rhaneuver; however, he was well aware of the surrounding terrain features and therefore would not be expected to attempt such a maneuver by losing altitude. - ~17- Attitude Inclicator,--Another explanation for the unwanted descent could be mechanical failure of the attltude indicator or failure of the pilot to maintain a positive climb indication by use of the attitude gyro, Attitude indicator gyros can exhibit precession as a result of neceleration forces. The typical precession error during longitudinal (forward) acceleration will produce a false nose-high indication momentarily. However, the allowable tolerance for a properly functioning gyro of the type installed in N41070 is so small that precession error should not have been sufficient to have caused noticeable effects unless there wes a malfunction. There was no evidence to suggest such a malfunction, although the gyro could not be tested because of impact and fire damage. However, since the pilot maneuvered accurately to the VOR, the attitude indicator was probably operating properly. A mechanical failure precisely at the time of the missed approach is highly improbable.

[ANALYSIS] There was no evidence to suggest such a malfunction, although the gyro could not be tested because of impact and fire damage. However, since the pilot maneuvered accurately to the VOR, the attitude indicator was probably operating properly. A mechanical failure precisely at the time of the missed approach is highly improbable. Therefore, the Safety Board believes mechanical failure of the attitude indicator probably was not a factor in the accident. Pilot_workload.--Several other problems could have occurred and several conditions could have affected adversely the pilot's ability to maintain a olimb attitude, and conversely could have allowed an unwanted descent to oecur. The pilot's instrument cross-check and visual seen would have included’ et least the following: attitude indicator, heading Indicator, vertical speed Indicator, altimeter, engine instruments (setting climb power » flap handle, gear handle, and possibly outside cues such as airport lights or other objects. All of these items would have had to be scanned during the few seconds immediately after the missed approach was initiated. The workload on the pilot would have been extremely high during this short period and would have required precise and timely perception, decisionmaking, and actions to effect a proper missed approach. Any distraction, such as the passenger's scream about an airplane on the runway, any misperception of cues, or any delayed action under these cireumstances could have caused an unwanted descent that was not detected in time to évcid ground contact. The rate of descent required to cause the accident is rather high and should have been readily Apparent to the pilots however, it could have gone undetected for a few seconds, which would have been sufficient to cause the accident. Physiological factors, --Sensory illusions, vertigo, or spatial disorlentation can cause a pilot to not detect a deviation, to interpret cues incorrectly, to take inappropriate action, or to take no action when action fs required. Any of these conditions can occur during instrument flight; however, the possibility of them occurring Is greater at night, during aircraft configuration changes, and when accelerations (aireraft maneuvering) occur, i.e. turns, climbs, and power changes. Also, the transition from instrument references to visual references and back to instrument references can induce illusions or disorlentation. Moreover, if the pilot looked out the left wiadow in reaction to the passenger's screain about an airplane on the runway or because of his own sighting of the ground or other features, the head and eye movemerts associated with this action could nduce disorientation, The ¢nly way for a pilot to overcome sensory Illusions {{s to monitor continuously and to rely on the attitude indicator, and to cross-check the other flight instruments, including the altimeter ar. vertical speed indicator, to verify that the aircraft is performing as desired. Unfortunately, when a pilot Is experiencing !llustonary events, the pilot may not detect an improper Instrument indication because the pliot’s Sensory organs are sending false messages to the brain.

[ANALYSIS] Unfortunately, when a pilot Is experiencing !llustonary events, the pilot may not detect an improper Instrument indication because the pliot’s Sensory organs are sending false messages to the brain. A pilot might réact to these false messages and take Inappropriate actions until the pilot determines from the airplane Instruments the need for correct actions. These actions require time, so any distraction or other reason for a breakdown of a pilot's Instrument scan could allow a deviation to oceur and to go uncorrected for a sufficient length of time to cause an accident. All of ~18- the known conditlons to induce illusions, vertigo, or spatial disorientation existed during Flight 104's attempted missed approach. However, the absence of conclusive evidence precludes the Safety Board from assigning physiological factors as causal. Although the preceding analysis does not include all of the possibilities for the cause of this accident, the Safety Board believes that the possibilities discussed are the more likely and plausible explanations for the aceldent. To assign any one of them as causal would be purely speculative and would not be supported by the evidence. Although the Safety Board could not determine the cause of this accident, it does believe that the low celling and reduced visibility were factors in whatever the cau 3 was. 2.3. Single-Pilot Operation A prior study 8/ conducted by the Safety Board and the evidence developed during this investigation have given the Safety Board concern about the adequacy of the regulations that allowed single-pilot IFR operation of Flight 104. Single-pilot operation in environmental conditions such a3 those existing at the time of the accident, coupled with the workload in conducting a nonprecision approach and missed approach in a twin-engine aleplane, is very demanding. While numerous successful Single-pilot operations occur dally in poor weather and high workload situations, the margin for error {{s much less during such operations because of the lack of redundancy provided by a second pilot. The issue of single-pilot IFR operations in commuter service was examined by the Safety Board as part of its special study of commuter airlines in 1980. Seventy percent of the operators surveyed as part of that study stated that their companies were authorized to conduct single-pilot IFR flights; however, many commented that the practice was "marginally safe" for many reasons. Among the reasons cited were the high workload factors associated with high-density ATC areas and alrport environments, and the demands of the cockpit which can overburden a single pilot. As a result of the commuter special study, the Safety Board made several safety recommendations to the PAA, some of which addressed the need to upgrade pilot experience and training requirements in general for commute: operations.

[CONCLUSIONS] 3. CONCLUSIONS 1. 2. The flighterew was certificated and qualified for the flight in accordance with company procedures and Federal regulations. There was no evidence that either crewmember was sufferin: from chronic or acute fatigue. There was no evidence of medical factors which would have detracted adversely from the physical ability of the captain to fly tr.> airplane. The first officer may have suffered from a dormant ear infection whieh could have detracted from his overall ability to perform his duties. Otherwise, there was no evidence of medical factors which would have detracted adversely from his physical ability to fly the airplane. The airplane was airworthy and maintained in accordance with the company's approved maintenance program. There was no evidence of a failure or malfunction of the airplane, its systems, or its components. However, there was insufficient evidence to discard the possibility of a failure, malfunction or mis-setting of a component of the navigation equipment. The surface weather reports and forecasts were current and substantially correct. The SAWRS facility at the Auburn-Lewiston Airport had not received a current inspection by the NWS at the time of the accident. The company's station agent was qualified to make weather observations and the SAWRS facility was in satisfactory condition at the time of the accident. The approach end of runway 4 lacked lights or other aids to more accurately determine the visibility. The visibility minimum on the instrument approach chart was published incorrectly. The ground-based navigational aids and approach light system were functioning properly and within acceptable tolerances. The combination of a shallow descent at high airspeed resulted in the airplane intercepting the glideslupe sooner than the flighterew expected and passing through glidepath. The flighterew may not have realized how close they were to the final approach fix and failed to configure the airplane for the approach in a timely manner and failed to follow established company procedures. ~35- The radar eontroller used poor judgment and technique when he attempted to assist the flight In intercepting the final approach course. The captain's acceptance of the large heading change issued by the radar controller contributed to the unstabilized approach. The captain attempted to complete an unstabilized approach and allowed the airplane to clescend below the GS. The altineters may have been mis-set at the time the airplane passed the final approach fix and the captain's altimeter could have been a factor in premature descent below DH. During the final seconds of the approach, the flighterew was probably infivenced by visual illusions caused by the weather, sloping terrain, high rate of descent, and darkness which contributed to e premature descent below DH.

[ANALYSIS] A witness observed the airplane leve}} off. However, the physical eviderice shows that the airpiene was in an average 6° descent path before and after initial tree impact. ‘The airvlane would have had to turn at least 20° to 30° to the left to have completed the approach. When it hit the trees, the airplane was correcting back to the runway and had obtained e ground track of 845° It is possible that the crew saw the approach lights when tl.ey descended below the GS. The physical evidence indicates that the landing lights were off, which otherwise could have obscured their vision in fog. Based on ths foregoing sequence of events, the Safety Board believes that the {{lighterew may not have been coimpletely aware of their position with respeet to the LOM until the controller attempted to assist the flight with the 340° heading. Although, the alrplane only turned 47° to the left instead of 60°, the turn given by the controller may have contributed to the ung’ abiliv. d approach. The flight was off course to such an extent that a 60° turn was require’ (uv put the airplane on a track to the LOM and a nonstandard procedure was required to achieve this objective. In view of this fact, the Safety Board believes that the controller used poor judgment and technique in his attempt to assist the flight. He also exereised poor judgment and technique in his subsequent handling of PEX 391 while it was on an ILS approach near the final approach fix. He kept PEX 391 too high and on a 52° intercept heading, which resulted in the airplane flying above the glideslope path. Since the crew was without gildepath guidance, they had to request a turn in order to position the airplane on course and on glidepath before commencing their final approach. The correet course of action in the case of flight 1808 would have been for the controller to have asked the pilot his intentions after providing the position report and then offered to provide radar vectors. Such action by the controller probably would have provided the captain with an easy aiternative. Acceptance of radar vector assistance would have alleviated his predicament and would have provided the crew an opportunity to establish a stabilized approach. However, the Safety Board must point out that the opportunity for the captain to have asked for radar vectors was always available. Instead, the approach was continued and the airplane was never established on the localizer or GS. Although the crew might have been able to make a landing, it would have been unsafe.

[ANALYSIS] Dueing the Safety Board's public hearing on the accident, the controller stated, ".. . this was just a turn, a correction to assist the pilot to intercept Lewie, south of, outside of Lewie."" The Board noted that, although the flight was in radar contact with Portland Approach, it had not been given radar vectors to the final approach course nor were they required. Instead, based on tneir previous clearance, the flightcrew was providing their own navigation to the final approach course by way of the transition depicted by the navigational aids on the instrument approach chart. The controller's aupervisor testified that the radar arrival section of the ATC Handbook (7110,65D) did not apply to flight 1808 because the flight was a nonradar arrival and the intercept angele timits specified in the radar arrival section did not apply to N300WP. The nonradar chapter of the handbook did not contain any criteria for handling aircraft in radar contact. The intent of paragraph 5-121, Section 9, "Radar Arrivals" is to provide a smooth transition for an aircraft from a radar vector to a fina) approach course without overshooting. However, it cid not provide guidance to controllers for use in assisting a flight when it deviates from a final approach course. In addition, the key distinction between a radar arrival and a nonrader arrival as presently defined, is whether or not an aiveraft is given radar vectors. The term nonrader errival is confusing because an aircraft would never be given radar vectors in a nonradar environment. In an accident involving Trans World Airlines at Berryville, Virginia, on December 1, t974, 9/ the controller issued an approach clearance withovt an altitude restriction and the airplane struck « hill during its descent while the crew was on their own navigation. FAA maintained that a pilot on final approach or transition to final approach, providing his own navieation while in radar contact, but not given radar vectors, was a nonradar arrival. (At that time, the radar arrival section, in the ATC Handbook (7110.8D), addressed the procedures of radar control for arriving aircraft, but the handbook did not define the term "radar arrival." The term, "non-radar arrival" did not exist in the manual] at that time.) $7 Aireraft Accident Report--"Trans World Airlines, Ine., Boaing 727-231, N54328, Berryville, Virginia, December 1, 1974" (NTSB-AAR~75~16), ~41- In its final report on the Berryville, Virginia, accident, the Safety Board stated that the flight should have been classified and handled as a radar arrival. ATC had radar contact with the flight, and there were no procedures in the manual on how to handle the flight using nonradar procedures when, in fact, the airplane was in radar contact.

[ANALYSIS] Acceptance of radar vector assistance would have alleviated his predicament and would have provided the crew an opportunity to establish a stabilized approach. However, the Safety Board must point out that the opportunity for the captain to have asked for radar vectors was always available. Instead, the approach was continued and the airplane was never established on the localizer or GS. Although the crew might have been able to make a landing, it would have been unsafe. In darkness and under the existing weather conditions, it would have been difficult for the flighterew to have recognized an excessive rate of descent until the airplane was very low. In addition, the flightcrew probably was susceptible to visual illusions because there were no other ground lights for reference in the appr: ich zone, there was water on the windshield, and there was a slight upslope to the runway. These conditions would have combined to produce a perception that the airplane was higher than it actually was in relation to the runway. To make a visual approach, a pilot must constantly interpret the changing visual scene to remain oriented on the proper glidepath to the runway. Pilots must judge height, speed, distance, and glidepath based on perceived movement of external objects, their brightness, and their apparent size as well as their relationship to the horizon and with other objects, and with respect to the airplane's orientation in space or its angular relationship with the runway. Thus, visual perception is a decision-making process based on a multitude of visual cues, At night, because visual cues are reduced and are further degraded by weather and sloping terrain, pilots must rely to a greater extent on flight and navigation instruments to reach the runway safely. The reduced number of visual cues, such as the -f SI. OR ADOEN ener e Te shh EA makin cual Saicobiel basic t diay aaa runway and approach lights, mey well provide images that ean be interpreted inaccurately, causing pilots to make inappropriate decisions. In this instance, flight 1808 could have become visual at about 1 to 11/2 miles from the airport. Therefore, the crewmembers would only have seen the 3,000-foot long approach light system and perhaps the runway threshold. Without further foveal or peripheral visual cues, an {{illusion would bs created that the airplane was too high in relation to the expected visual light scene, i.@., 3,000 feet of approech lights plus 5,000 feet of runway lights. An additional visual iflusion would have been creaced by the featureless and unlit terrain surrounding and preceding the approach lighting system. This is called the "black-hole approach” and concerns approach to a lighted runway (or in this case to the approach lighting svstem) over dark featureless terrain beyond which the horizon is not discernible. Simulated tests of both visual conditions have shown that a pilot tends to descend along the are of a eirele, the circumference of which touches the ground short of the runway lights.

[ANALYSIS] This is called the "black-hole approach” and concerns approach to a lighted runway (or in this case to the approach lighting svstem) over dark featureless terrain beyond which the horizon is not discernible. Simulated tests of both visual conditions have shown that a pilot tends to descend along the are of a eirele, the circumference of which touches the ground short of the runway lights. The fact that flight 1808 had water on the windshield and that the airplane was manuevering and was not aligned with the runway and approach lights center line aggravated the situation. 10/ Therefore, the Safety Board believes that the captain uset poor judgment by not making a missed approach in the vieinity of the LOM. Once he decided to continue and attempted to "salvage" the approach, the crew had difficulty because the airplane was not only unstabilized on the ILS, but it was outside the LOC and GS limits. This degraded significantly their ability to determine the airplane's correct angular relationship to the runway when the runway environment came into view. The Safety Board believes that the visual and meteorological conditions could have contributed to the pilot flying too low. Even with the VASI installed, the crew probably did not see it and if thev did, the system does not indicate how far an airplane is above or below the proper glidepath. On site examination showed that the landing gear was extended with the flaps retracted instead of at the approach or full down position. There are four possible explanations. The crew may have accidentally hit the flap lever after the airplane struck the trees, they may have failed to extend the flaps because they were hurried, or they may have failed to complete or properly coordinate a missed approach procedure. If they attempted to execute a missed approach, the proper procedure in the Beech 99 requires first arresting the rate of descent by increasing’ the pitch attitude and adding full power, and after a positive rate of climb has been established, retracting the landing gear, and then retracting the flaps. The captain must raise the landing gear, and the first officer raises the flaps. If the crew had attempted a missed approach, the change in control yoke pressure should not have been significant if the airplane had been trimmed at a speed between 120 and 130 knots. The position of the ctabilizer actuator suggests that this would have been the case. About 130 knots, the airplane has the capability of generating sufficient acditional lift in a 2 G pullup to arrest the 1,300 fpm rate of descent within about 10 feet. Also, at this speed the different flap position effeet is negligible. Therefore, based on a 3-second recognition and response time and a 1-second aircraft response time, the pilot could have arrested the vate of descent in about 80 feet. Although remote, a fourth possibility is a mechanical malfunction or electrical short in the system.

[ANALYSIS] If the airplane was tracking outbound on the 036° radial at 26 miles, it actually would have been about 0.7 mile to the right of the desired course. At 2159:50, flight 1808 was actually 1.3 miles to the right of the 036° radial at Doley, about 26 miles away. The ‘.d-mile difference between the flight's position and the 036° radial would have resulted in a course deviation indicetor (CD) needle deflection of only seven-tenths of a dot to the left (fly left indieation). The CDE would have indicated that the airplane was nearly on the 036° radial. At that time, the alrplane made a 10° heading change to the let't. 4 ¥ ERE AEA, EPATE AAD gas MAIR AED ATURE LIES END MEY MEL RGF LCIEE CRISES EURO MS A al Wt Ht, 4 eas entr tyson ool meine - ~42- The remainder of the approach to the Doley intersection, where the 041 localizer course would have been intercepted, was flown erratically. A constant rate of descent was not established to cross Doley at the desired 3,000 feet. The crew should have realized that in 26 miles from ENE to Doley, they had to lose 4,000 feet «vhich, at an airspeed of 200 to 210 KIAS, would require a rate of descent of 550 to 600 feet per minute. Instead, the airplane was initially established at a rate of descent of 500 feet per minute and then was allowed to decay to 350 feet per minute. The rate of descent then increased to 800 feet per minute when the crew was told that they were 12 miles south of Lewle and cleared for the ILS approach. It is not known why, at this point, the airplane began a ground track 6° to the right, away from the track that would have permitted intercepting the final approach course. It could be attributed to inattention or misinterpretation of displayed navigational information. Also, 4 signal anomally, malfunction, or incorrect setting of the NAV equipment could all be possible explanations for the turn away from Doley and the final approach course. If the NAV equipment had been properly ‘uned to and receiving the ILS frequency after passing Doley, the GS needle should have started moving downward toward the "bullseye." The distance measuring equipment (DME) reading from ENE should have indicated about 30 miles at this location, in addition to the CDI needle movement if one NAV had been tuned to Navy Brunswick.

[ANALYSIS] It is, therefore, eoneeivable that in view of the noncomplexity of the airplane and the brevity and simplicity of the checklist, checklist items probably were performed silently by the nonflying pilot. Performing checklist items in this manner in a single pilot operation is obvious, but carrying such habits over into a two-pilot air carrier operation fosters apathy, negates the safety aspect of crew redundancy, defeats the purpose of the crew concept, and invites mistakes that go unnoticed. Consequentiy, the Safety Board believes that a lack of crew coordination could explain, in part, the operational discrepancies in this accident. The ATC tape of communications revealed that both the captain and first officer were making the radio transmissions, contrary to corpany procedures, In part, for this reason, the Borrd was unable to determine who was flying the airplane. The evidence of how the radio communieations were handled by the erew is Indicative of poor crew coordination technique and could have resulted in a distraction to the flying pilot. Therefore, the FAA must use caution in its certification and surveillance of commuter airlines to ensure that crew coordination is practiced and it should emphmusize the problems that can develop with airplanes that are designed for the single pilot. - ect gan ae RIPEN DY LRRD LPI Lt ONES ELE LOE LCE RO ERTS BOT EEN Cok s One such unnoticed mistake that could have evolved from the unchallenged checklist use, is a mis-set altimeter. Both altimeters were found set at pressure settings that were different than what was provided by the station agent. The captain's altimeter was set at 30.42 and the first officer's was set at 30.12, The station agent stated that he had provided the crew with an aitimeter setting of 30.24. The last altimeter setting brovided by Portland Approach was 30.26 which was acknowledged and repeated by the irst officer. Based on the manufacturer's examinations, the captain's a:iimeter setting could have been significant since its reading would have been indicating an altitude close to DH when the airplane struck the trees. This would have misled the captain to believe that the airplane was higher than its actual altitude. However, the copilot's altimeter setting was mis-set in the opposite direction. It is possible that the disparate altimeter settings were caused by ground impact and subsequent destruction of the cockpit. However, it only remains a possibility since the altimeters showed no evidence of barometric gear system derangement or impact marks to fix the position of the altimeter settings at the time of the accident. From a human performance standpoint, it is possible that the captain may have mentally transposed the last two digits of the altimeter setting when he heard the transmission and transferred 30.42 instead of 30.24 to his altimeter. This would require 1 3/4 turns with the thumb and forefinger.

[ANALYSIS] The first officer had no known history of ear problems and apparently he did not have a ecid. The evidence does not permit a determination that the first officer suffered from the affects of aerotitus media (the inability of the middle ear to ventilate itself) resulting in inflammation of the membrane and surrounding tissue. Since knowledge of the first officer's complaint was learned after the autopsy was performed, the Safety Board was not able to request the medical examiner to specifically investigate this possibility. Nevertheless, a dormant infection or other latent cause of tissue swelling eannot be dismissed entirely. This condition conceivably could have caused a sudden onset of severe pain in the ear resulting in a distraction to the first officer and a disruption in crew coordination and in the overall performance of the flight. 2.3 Airplane Maintenance The investigation found no evidence of a failure or malfunction of the airplane, its systems, or its eemponents. The history of problems with the NAV/COMM equipment was considered as a potential factor in the accident. Water leaks could have affected the NAV/COMMs and could have caused erroneous navigation displays resulting in an unstabilized flightpath. Although the No. 1 NAV/COMM removed from the airplane the day before the accident showed no evidence of water contamination, and there was no further writeup concerning water leaks, it could be concluded that the water leak problem had been corrected when the windshield was last repaired on May 21, 1985. However, another NAV/COMM discrepancy due to a water leak through the windshield about 2 weeks later shows that the last repair did not correct the water leak problem in N300WP. The discrepancy and corrective action history shows there was continuous problems with the NAV/COMiM. Thus, a remote possibility of some form of malfunction remains because of the location of the avionies and other electrical components in the nose compartment of the airplane adjacent to the windshield, anc because the airplane had been flying through light rain at the time of the accident. 2.4 Weather Reporting System The SAWRS facility at the Auburn-Lewiston Airport was found unsatisfactory during an inspection on April 11, 1983; however, a reinspection of the facility 2 days after the accident showed that the previous discrepancies had been corrected and that it was in satisfactory condition at the time of the accident.

[CONCLUSIONS] The altineters may have been mis-set at the time the airplane passed the final approach fix and the captain's altimeter could have been a factor in premature descent below DH. During the final seconds of the approach, the flighterew was probably infivenced by visual illusions caused by the weather, sloping terrain, high rate of descent, and darkness which contributed to e premature descent below DH. The cockpit noise level in the Beech 99 probably was a fector in the flighterew's ability to communicate with each other. Compsny management failed to insure that flighterews were adhering to company standards and operating procedures. FAA surveillance was affected adversely by inadequate manpower and failed to insure that operational standards within the company were maintained. 23. The airline’s pilot turnover rate was sufficient to have reduced the overall experience level of the pilot workforce within the company. 24, The accident was not survivable. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of the aceident was the captain's continuation of an unstabilized approach which resulted in a descent below glideslope. Contributing to the unstabilized approach was the radar controller's issuance and the captain's acceptance of a nonstandard air traffie control radar vector resulting in an excessive intercept with the localizer. 4, RECOMMENDATIONS As a result of its investigation, the Safety Board reiterated the following recommendations to the Federal Aviation Administration: Establish for aircraft used in commercial operation the maximum cockpit noise levels which will permit adequate direct volce communication between flight crewmembers under all operating conditions. (Class Il, Priority Action) (A-81-75) ryt ~ i? we r) t H ber sO . . . ‘ i? gs ROR ante cpanel eetin e er anY Ele ERE Set ce rintoe carenn eee: ~58- Require the installation and use of crew interphone systems in the cockpits of those aireraft in which noise levels reach or exceed the maximum level established for adequate direct voice communication between flight crewmembers under all operating conditions. (Class II, Priority Action) (A-81-78) Require that all multiengine, turbine-powered, fixed-wing aircraft certificated to carry six or more passengers manufactured on or after a specified date, in any type of operation not currently required by 14 CFR 121.343, 122.359, and 135.151 to have a cockpit voice recorder and/or a flight data recorder, be prewired to accept a "general aviation" cockpit voice recorder (if also certificated for two-pilot operation) with at least one channel for voice communications transmitted from or received in the aireraft by radio, and one channel for audio signals from a cockpit area rnierophone, and a "general aviation" flight data recorder to record sufficient data parameters to determine the information in Table las a function of time. (A-82-107) Require that "general aviation" cockpit voice recorders (on aireraft certificated for two-pilot operation) and flight data recorders be installed when they become commercially available as standard equipment in all multiengine, turbine-powered fixed-wing aircraft and rotoreraft certificated to carry six or more passengers manufactured on or after a specified date, in any type of operation not currently required by 14 CFR 121.343, 191.359, 135.151, and 127.127 to have a cockpit voice recorder and/or a flight data recorder. (A~82-109) Require that "general aviation" cockpit voice recorders be installed as soon as they are commercially available in all multiengine, turbine-powered aircraft (both airplanes and rotorcraft), which are currently in service, which are certificated to carry six or more passengers and which are required by their certificate to have two pilots, in any type of operation not currently required by 14 CFR 121.359, 135.151, and 127.127 to have a cockpit voice recorder.

[ANALYSIS] ATC had radar contact with the flight, and there were no procedures in the manual on how to handle the flight using nonradar procedures when, in fact, the airplane was in radar contact. The Board recommended on July 24, 1975, that the PAA: A-15-58 Define the term "radar arrivel" and assign an equal weight of controller responsibility to all arrivals receiving radar service, regurdiess of the kind of radar service. The FAA responded on August 18, .975, that it concurred with the recommendation, that the recommendation was part of a study to review terms and phrases used in the ATC system for the purpose of determining whether they needed to be defined, and that it would cake whatever corrective action would be needed. As a result, the FAA included the pilot/controlier glossary of ATC terms in both the ATC handbook and the Airman's Inforrnation Manual. The Safety Board believes, however, that the definitions are confusing and somewhat inconsistent with the environment under which an aircraft may be operating. The Board believes that a standard should exist for intercept angles regardless of whether or not an aircraft: is receiving radar vectors and that appropriate material should be included in the handbook that :ddresses the situation when an aircraft deviates from its course in a radar environment. If the aircraft eannot be returned to course using the standard criteria for an intercept, then the pilot should be informed that he appears to be too far right or left of course for a safe approach, and asked his intentions. An aircraft in radar contact should not be handled under a chapter af the handbook that was designed and intended for nonradar operations. In the case of flight 1808, the off-course flightpath was identified by using radar, the correction issued to the pilot was based on radar iaformation, and the position report given to the crew with respect to Lewie was accomplished using radar. Therefore, the Safety Board believes that the flight should have beer treated as a radar arrival. The procedures for handling flights should be those governing radar operations and not those governing nonradar operations. To accomplish this, the definition for radar arrivais should be amended te include all IFR arrivals under radar contact, and the definition of nonradar arrivals should be armended to include only arrival aireraft not in radar contact, If deviations are noted, the procedures for assisting a flight to return to course using radar should be added to the radar arrival section in the chapter titled, "Radar." 2.7 The Fuizhtpath A review of ali available data pertaining to the navigational aids and the radar track of the airplane disclosed a possible explanation for the slight off~course track after flight 1808 passed ENE, The ENE VOR was misaligned by 1° to the right and the 036° radial had a 2.5° right bend at 20 miles.

[ANALYSIS] It may be chat verification of runway heading is such a rudimentary procedure that the air ecezrier believed that specialized training was not necessary. While such a belief may have been reasonable and reflective of accepted practice, that this crew failed to carry cut this basic step indicates that a deficiency which nedds. to be addressed may exist in al cerrier crew training’ and certification procedures. : I : The Safety Board cannot explain why the captein of KAL 084 devided to take off in the fase of his uncertainty as to whether his airplane was holding at runway 32, The captain stated: I I : ~21- -.+T felt unsure that the aircraft was on the correct runway....I discussed this with my copilot [the first officer}} who felt sure that we were on the correct runway. After 3-4 minutes of discussion, I considered teking runway 6R because of my uncertainty. However, the runway size and lighting appeared to be correct so I decided to take off. This statement indicates that the captain failed to recognize that his familiarity with the airport would not compensate for the limitations in other sourees of information he would use ordinarily to confirm the aireraft's location. The captain failed to exercise proper decisionmaking responsibility by relyings too heavily on the first officer's belief that the airplane was on the correct runway. Proper command procedures should hive dictated to the captain not to commence takeoff without confirming that he was holding at runway 32. The captain's statement indicates that he felt that the first officer, who had a higher level of recent experience at the airport than the captein, was more certain about the aircraft's location than the captain was. The first officer stated that, "In spite of poor visibility, our aircraft advanced and was able to get onto the east~west taxiway." The evidence indicates that KAI 084 was never on the east-west taxiway. Unlike the captain, the first officer in his statement did not manifest ary uncertainty about the nircraft's location. The Safety Board believes that the first officer's strorg belief about their location may have influenced the captain's decision to commence takeoff. The first officer's confidence regarding being on the correct runway in the face of the csptain's uncertainties constituted a slight role reversal in that the captain's overall command authority when deciding te take off was influenced by the first officer's comments. In the past, the Safety Board has encouraged assertiveness training for first officers, to exercise their responsibilities as part of the cockpit team; however, a companion responsibility for captains to exercise positive cockpit crew management must exist. In this instance, the crew concept broke down. This breakdown may have been due to the crew's intense eoncentration on the airport surface markings and runway and taxiway signs in order to confirm their location.

[ANALYSIS] The captain stated that while taxfing, he attempted to keep the airplane centered on the yellow taxi line but because of snow and ive gr nd cover and the reduced visibility, he could not Powitively identity his location on the airport onee the airplane left the cargo ramp. The captain stated that he turned the aircraft right from the north apron to what he and the first officer believed was the east-west texiway. The Safety Board believes that the crew actually turned, not about 100° to the right which would have turned them onto the esst-west taxiway, but about 60° right oy to taxiway W~1. From there, the captain lates turned the airplane ebout 50° right, instead of about 90°, to what the flightcrew believed was runway 32, but to what was, in fact, runway w4R. The taxi tests Strongly suppor: this as the most likely taxi route. , Because of the large size of the DC-10-90, which may distort the pilot's sense cf motion and the restricied surface yiaibility, the Safety Board believes thet the captain of KAL 084 could have experienced difficulty in distinguishing batween the turn of &0° instead of 100° or the turn of 50° instead of 90° while taxiing slowly and Straliiing to see the taxiway and runway markings, since outside visual cues were limited. {{t is diftteult to uriderstand, however, why the captain and first officer, foliowing some discussion, about runway uncertainty, did not use their directional gyros or the standby compass to orient themselves with regard to headings, especially after they had aligned the airplane with what they believed was runway $2 and had discussed it for 2 to 3 minutes. If any ont of the flighterew had checked the heading indicators, it should have been apparent before ti takeoff roll that the airplane was positioned on the wrong runway. The KAL checklist dis not require a pretakeoff heading shacks however, other airline checklists require protakecff runway confirmation and accepted practice is ts check heading indicators before starting takeoff. : The first officer's statement concerning the sighting of SCA 59--"when the aireraft's speed reached about V1, I caught sight of a small aircraft about 15 metors in front of us..."~-was most likely inaccurate and may further indicate some degree of misperception on his part. The airspeed would have been about 100 knots at that point, well below V1, and the accurate judgment of distances in those circumstances would have been difficult. These commenis were another indication of misperception by the first officer. I The primary sources of information that are ordinarily available to crewmembers for guicance on airport surfaces were either partially or completely unavailable to the crew of KAL 084. At nighttime or under limited visibility conditions, crewmembers rely on runway surface markings such as taxiway lines and runway numbers, taxiway and runway lights, and runway and taxiway signs to provide them with information concerning their location on the airport.

[FINDINGS] Flighterews must be especially vigilant during taxi, hold, and takeoff operations and must make extraordinary efforts if needed to stay aware of their position on the airport at all times. Crew coordination procedures should be enhanced and particular alertness should be practiced when visibility is reduced by inclement weather. I }} 3. CONCLUSIONS Findings 1. Both airplanes were certificated and maintained in accordance with approved procedures. 2. There was no evidence of preaccident failure or malfunction of either airplane's structures, systems, powerplants, or flight control systems. 3. The pilot of SouthCentral Air Flight 59 (SCA 59) was properly certificated and qualified for this scheduled domestic passenger flight. His actions did not contribute to the accident. The flighterew of Korean Air Lines Flight 084 (KAL 084) were properly certificated and qualified for this scheduled cargo flight. The flighterew of both eirplanes irvolved held current medical certificates. Both the captain and the first officer of KAL 084 had extensive experience operating into and out of the Anchorage International Airport, which should have reduced the ; robability of crew disorientation while taxling in the low-visibility conditions. The decision of KAL 084's captain to use runway 32 for departure was not in accordance with KAL operating specifications. -~34~ The obscuration of runway and taxiway markings eat the airport adversely affected the performance of the flighterew of KAL 084 by causing them to give disproportionate attention to locating the runway markings. The most likely taxi route, taken in error, by KAU 684 was south along the west side of the north apron, right ontv taxiway W~-1, and right again onto runway 24k. The flighterew of KAL 084 could have determined that thsir airplane was lined up on the wrong runway if they had cross-checked their heading indicators. Based on the estimated takeoff gross weight of KAL 084, the runway length required for takeoff was 8,150 feet. Since the actual length avaliable to KAL 084 on runway 24R was about 2,400 feet, an accident would have resulted even if KAL 084 had not collided with SCA 59. . By raising the nose of his airplane and turning his sirplane slightly to the left, the captain of KAL 084 avoided inflicting extensive damage to the fuselage of SCA59 and probable fatal injuries to the crews and passengers onboard both airplanes as a result of the collision. Of the four runway and taxiway signs KAL 084 would have passed on the most likely taxi route it erroneously took, one had no illumination, one was oniy partially illuminated, and two were fully illuminated. There was no taxiway guidance sign at the intersection of taxiway W-1 and the east-west taxiway. Operators of airports certificated under 14 CFR Part 139 are not required to place standardized signs at each taxiway/runway and taxiway intersection.

[FINDINGS] There was no taxiway guidance sign at the intersection of taxiway W-1 and the east-west taxiway. Operators of airports certificated under 14 CFR Part 139 are not required to place standardized signs at each taxiway/runway and taxiway intersection. Runway signs should be sufficiently different in design from taxiway signs so thet they alert the operators of ail surface vehicles and airplanes of the nature of the intersection. 17. Lighted runway/taxiway signs should be inspected daily to ensure their operability and maintained as required. 3.2 Probable Cause The National Transportation Safety Board determines that the probable causes of ths accident were the failure of the pilot of Korean Air Lines Flight 084 to follow accepted procedures during taxi, which caused him to become disoriented while selacting the runway; the failure of the pilot to use the compass to confirin his position; and the aceision of the pilot to take off when he was unsure that the aireraft was positioned on the correct runway. Contributing to the accident was the fog, which reduced visibility to a point that the pilot could not ascertain his position visually and the control tower personnel could not assist the pilet. Also contributing to the accident was a lack of gible taxiway and runway signs at several intersections passed by Flight 084 while it was a oe SNH A ROPE 9 “ SORRY BENE LE AER se SEES gy I ERRATA ET Rp cor Fete ae: eee seer fe vad . aed ‘ f > 4. RECOMMENDATIONS I As 4 result of this accident investigation, the National Transportation Safety Board recommended that the Federal Aviation Administration: Require that airports certificated for air carrier operations install signs at all runway and taxiway entrances, exits, and intersections that indicate the identity of the runway or taxiway. (Class ii, Priority Action) (A-84-98) Require that the graphics on taxi:vay/runway identification signa be standardized and of sufficient size to enable them to be legible to aircraft crewmembers in all meteorological conditions in which air carrier operations are authorized. (Class i, Priority Action) (A~-84--99) Require that airport operators inspect and tmatntain the lights illuminating airport taxiway/runway identification sigs as part of the cay rita inupection requizements. (Cless I, Friority Action) A-84-100 Require at all airports certificated for air carrier operations that uniform signs be installed which are classified by function (e.g., runway entrance, runway exit, taxiway intersection) with each function having a unique shape, color, end/or size so that runway entrance signs are distinguishaole from all other advisory siens on airport property. (Class Ii, Priority Action) (A~84-101) Require that air carriers incorporate in training of their crewmembers procecures and responsibilities during ground operations in restricted visibility conditions, to enable them to operate sefely in such conditions. (Class I, Priority Action) (A-84-102) - BY THE NATIONAL TRANSPORTATION SAFETY BOARD — /s/ JIM BURNETT Chairman /s/ PATRICIA A, GOLDMAN Vice Chairman /s/ GH, PATRICK BURSLEY Member VERNON L, GROSE Member August 9, 1984 oh emedeal Fey OSU Se ne ara ed ie sere Preceding page blank = -27- APPENDIXES APPENDIX A INVESTIGATION AND HEARING

[ANALYSIS] As the captain of KAL 064 turned left to miss SCA 59, the main body gear swung to the right and struck the left wing of SCA 59, knocking the wing off outboard of the engine nacelle. The left main gear of’ KAL 084 struck the right wing of SCA 59 in the area of the engine and sheared off the wing at the wing root, and continued back and struck the right horizontal stabilizer of SCA 59. As a result, while the wings und vertical stabilizer were separated, the fuselage of SCA 59 remained intact and the passengers suffered no serious injuries. I! the captain of KAL (84 had failed to take these actions, either the nose wear or the center main body gear of KAL 084, or both, might have struck the fusolage of the smaller airplane and probably would have resulted in fatalities aboard SCA 59 and possibly aboard KAL 084. 2.4 RAL 084 Hightcrew Operational Factors The KAL 084 crewmembers were experienced in operating the DC-10; the captain had logged over 6,000 hours tn the airplane and the first officer almost 3,006 hours. Additionally, the captain and first officer were experienced in operating at Anchorage International Airport. The captain had logged 73 landings and 78 takeoffs from the airport in an 8 1/2-year period prior to the accident, and the first officer had logged 68 landings and 66 takeoffs in a 3 3/4~year perio, : The captain's decision to use funway 32 for departure was not in accordance with KAL operating specifications, A prevailing visibility of 1/4 mile was requlred and the prevailing Visibility at the time of the accident was 1/8 mile. The RVR readings for runway 6R were above minimums and the capain should have selected that runway for departure. While the captain's decision did not directly bear on the sceident. since he attempted takeoff on a runway other than the runway to which he was cleared, it was an operational deficiency end indicates performance not in keeping with that expected of an air carrier captain. The Safety Board cannot determine precisely the procedures the KAJ. crew used while taxiing since the cockpit voice recorder wan not recovered. Anchorage ground control cleared KAL 084 to tart tio runway 32 and asked the flighterew to report entering the east-west taxiway. The captain stated that while taxfing, he attempted to keep the airplane centered on the yellow taxi line but because of snow and ive gr nd cover and the reduced visibility, he could not Powitively identity his location on the airport onee the airplane left the cargo ramp. The captain stated that he turned the aircraft right from the north apron to what he and the first officer believed was the east-west texiway.

[ANALYSIS] In the past, the Safety Board has encouraged assertiveness training for first officers, to exercise their responsibilities as part of the cockpit team; however, a companion responsibility for captains to exercise positive cockpit crew management must exist. In this instance, the crew concept broke down. This breakdown may have been due to the crew's intense eoncentration on the airport surface markings and runway and taxiway signs in order to confirm their location. The Board believes that such a situation may lead to a breakdown in carrying out individual cockpit responsibilities unless the wrawmembers have been traiixed to recognize and react to the situation. Because the crew of KAL 084 commenced takeoff in spite of the uncertainty regarding their location on the airport, the Safety Board is concerned that the crew was not properly trained in ground operations in marginal meteorological conditions existing at the time. A common procedure for takeoffs in restrieted visibility is for pilots to crosicheck their gyro/compass heading with the runway heading prior to commencing takeoff. Crews should be trained to perform such a procedure regardiess cf how selfevident their position may appear te them. Af a result of this accident and similar errors in air carrier ground operating procedures demonstrated by ground collision accidents at airports during restricted visibility conditions, as well as by the ASRS cata, the Safety Board is concerned that flighterews are not being adequately trained in maneging cockpit resources and coordinating their responsibilities when operating in marginal ground nianeuvering conditions that require intense conceatration. The need for specific training in ground operation procedures for crews is especially Important, since there are no requirements for Standardized, illuminated, and easy-to-read runway and taxiway signs at airports certificated for air carrier operations. When there ts xbscuration of taxiways and runways added to restricted visibility, the need for a crew that is well trained in ground operations becomes critical. It is not possible for alr traffic controllers during these conditions to. verify an aircraft's location on the airport, in the absence off a radar such as ASDE that tracks airport surface traffic, other than relying on the crew to accurately report their location. a sme arse ee a gL gers * . , . ea - ~ : z ¥ Laan : ae ESSER ETS ts 4 + ce - i. Oey ~ : _ . Logie ; . . ~ L at Tees =) re Se ee Si ree re rid . nee coe : . z : * et ‘ : — fet ee eee rer eres “ : . ae 2.6 Airport Signs and Environment The demands on the crew of KAL 084 while they were taxiing were not excessive for a highly experienced crew, despite the lack of much of the information that crews usually rely on to taxi caused by the Hmited visibility and absence of ASDE. The Safety Board examined several of the runway and taxiway signs at the airport to determine if all of the avaiiabie sources of ground location information external to the airplane were adequately presented to the KAL 084 crew.

[CONCLUSIONS] The Eastern flightcrew had three opportunities to learn about preceding landing traffic by listening to the tower frequencies; however, the time between transmissions, the large number of transmissions, and the required duries in the cockpit would have limited the utility of that information. Airspeed reduction transmissions to N44UE by the radar monitor controller were insufficient to achieve the required 4 miles separation from the preceding airplane, CO 9687, on the final approach and at the threshold. The absence of appropriate airspeed reduction instructions to EA 111 by the Atlanta approach north final and the radar monitor controllers led to a speed differential that resulted in a loss of the separation between EA 11] and N44Ué. The traffic volume at the time of the accident presented an average controller workload, but the local controller was distracted with radio difficulties (misunderstood instructions) when communicating with CO 9687. 42 The local controller’s distraction by communication difficulties with CO 9687 was prompted by his perceived need to clear runway 26 left for another airplane inbound with a hydraulic emergency and the possibility of a runway incursion from CJ 9687 during taxiing. The controller was inattentive to the more immediate task of monitoring the separation of traffic landing on runway 26 right. 3.2 Probable Cause The National Transportation Safety Board determines that the probable causes of this accident were (1) the failure of the Federal Aviation Administration to provide air traffic control procedures that adequately take into consideration human performance factors such as those which resulted in the failure of the north local controller to detect the developing conflict between N44UE and EA 111, and (2) the failure of the north local controller to ensure the separation of arriving aircraft which were using the same runway. Contributing to the accident was the failure of the north local controller to follow the prescribed procedure of issuing appropriate traffic information to EA 111, and failure of the north final controller and the radar monitor controller to issue timely speed reductions to maintain adequate separation between aircraft on final approach. 4. RECOMMENDATIONS Therefore, as a result of the investigation of this accident, the National Transportation Safety Board recommends that the Federal Aviation Administration: Develop an Air Traffic Bulletin and provide a mandatory formal briefing to all air traffic controllers on the importance of, and the need for, giving traffic information when issuing an anticipated separation landing clearance. (Class II, Priority Action) (A-91-27) Amend the Air Traffic Control Handbook, 7110.65F, paragraph 3-127, to preclude the issuance of multiple landing clearances to aircraft outside of the final approach fix.

[ANALYSIS] The CO 9687 fligh:crew took positive action to avoid becoming a hazard. They stopped clear of the active runways and remained in position until their clearance was clarified. 34 The Safety Board conciuded that there was a lack of understanding between the tocal controller and CO 9687, and that this communications anomaly did not result from any oquipment failure. Rather it was the result of an incomplete transfer of intormation (taxi instructions) between the controller and the flightcrew of CO 9867. The net effect of this lack of information transfer was to create a self-imposed workload on the controller that was sufficiently high to cause him to disregard other higher priority tasks. If the local controller had been so concerned that the airplane was going to cross runway 26 left without a clearance, he had the option of discontinuing departures from that runway and directing ASE 301, that was holding in the takeoff position, to clear runway 26 left. As long as ASE 301 was holding in position on the runway, it could not be threatened by a possible runway incursion from an airplane at the opposite end of the runway. However, the local controller became distracted for a critical period by the possibility of a runway incursion involving ASE 301 and CO 9687. Eventually, et 1904:13, the local controller cleared ASE 301 for takeoff on runway 26 left. The collision of €A !1] and N44UE had taken piace on runway 26 right at 1904:07. It is recoynized that concentration on one task can overload a person to the extent that other relevant cues are disregarded or otherwise not attended to, leading to a degradation of overall task performance. Because the north local controller focused his attention on the path of CO 9687 on taxiway Bravo, at the west end of runway 26 left, he was distracted at a critica) time from the landing rollout of N44UE and the FA 111 airplane that was about to cross the threshold and land on the same runway. The Safety Board concludes that this accident was a result of lapses in the performance of the Atlanta tower north tocal controller and, to a lesser extent, the performance of the Atlanta approach control north final controller and the radar monitor controller. Specifically, the north local controller did not ensure the separation of the aircraft approaching and landing on runway 26 right. further, ne failed to follow the prescribed procedure of issuing appropriate traffic information to the crew of EA 11). This information would have improved the flightcrew’s situational awareness and their motivation to search for the preceding King Air. The Atlanta approach north final controller and the radar monitor controller had opportunities to issue timely speed reductions to ensure that adequate separation was maintained between the successive aircraft on final approach, but did not do so. Although these lapses of controller performance are cited as causal, the Safety Board also has chosen to recognize that the controllers’ performance was a direct product of FAA air traffic management institutional decisions and practices that do not allow for human performance lapses in judgement or decision making.

[ANALYSIS] Weather conditions were at or above the criteria for visual flight rules. However, there were scattered clouds at 500 feet and fog present 29 about the time of the accident. In postaccident interviews, neither the flightcrew of EA 111, the Epps Air Service copilot, nor the air traffic controllers identified environmental factors as a constraint to the normal performance of their duties. Evidence was conclusive that the collision occurred = on runway 26 right about 3,000 feet from the approach end. There were aircraft tire scuff paths with measurements corresponding to the dimensions of a B-727 landing gear. There were small parts of the King Air airplane scattered about the right side of the runway at this point. The geometry of the collision was indicative of a similar heading, and longitudinal impact was consistent with the ground tracks of both airplanes along runway 26 right. In considering factors that interfered with visually identifying objects on the runway, the Safety Board considered the effects of lighting from a large automobile manufacturing complex to the east of the airport and the air cargo ramp adjacent to runway 26 right, as weil as lights from airport emergency vehicles. Although each of these light sources made the visual acquisition of aircraft more difficult, the Safety Board did not consider the singular or cumulative effects of these sources to be a significant factor. Therefore, other reasons were considered to explain why neither the EA Ill flightcrew nor tower personnel made visual contact with the King Air airplane on runway 26 right during the critical period preceding the collision. 2.2 Airplane Airworthiness and Conspicuity The operating condition of each of the involved airplanes was reviewed as possible contributors to the collision. The 8-727 was maintained in accordance with the applicable directives, and there was no evidence that any windshield deficiency or other malfunction contributed to the collision. The evidence indicates that the King Air was not in compliance with airworthiness requirements because of deficiencies in the anticollision lighting system. During the departure from DeKalb Peachtree Airport, an Epps Air Service flight line maintenance technician observed that the red anticollision light atop the vertical stabilizer and the white strobe light installed in the fuselage tailcone were inoperative. Further, the red anticollision light on the lower fuselage was not of an approved type and was not functional. The "Beacon" switcn that controls the red anticollision lights on both the top anu bottom of the airplane was reported to be inoperative by ground personnel prior to the flight. The lighting discrepancies were not entered in the aircraft log book and were not repaired before flight as required by FAA regulations. The Safety Board was not able to establish that the pilot-in-command was aware of these lighting deficiencies although lighting is a pert of the preflight checklist. Therefore, the deficiencies would become apparent during a properly conducted preflight exterior inspection. In normal operations where collision avoidance is dependent upon pilot see and avoidance, anticollision lighting plays ar important part in 30 airplane conspicuity at night. The operation of required lighting equipment is an obligation for al}} pilots and operators.

[ANALYSIS] No indication was given that they were number two for landing behind another airplane. The Safety Board believes that flightcrews are conditioned to receive such information. as required in the Air Traffic Con*rol Handbook procedures relating to anticipating separation. If the controller had provided traffic information to the EA 11] flightcrew, the flightcrew’s sense of situational awareness and motivation to search for a preceding airplane might have been increased. Lacking such information, it appears that the crew proceeded through their normal task of completing a routine night landing on a runway to which they had been cleared, unaware that there was another airplane on the runway. The fact that EA 111 had received a landing clearance did not relieve the flightcrew of responsibility to "see and avoid" other aircraft in their vicinity. However. in the absence of conspicuous lighting on the King Air and without prompting from ATC to direct their attention to traffic ahead, it was extremely difficult, if not impossible for the EA 11] flightcrew to detect the other aircraft on the runway. Moreover, there is a concept known as diffusion of responsibility that describes a tendency on the part of pilots in some circumstances to relax their vigilance. A National Aeronautics and Space Administration study on near midair collisions‘ sBillings, C., Greyson, ®., Hecht, W., and Curry, R., "A Study of Near Midair Collisions in U.S. Terminal Airspace," NASA Technicat Memorandum 81225, 1980. 32 indicates that an inappropriate sense of shared responsibility may occur when an airplane is under ATC radar control. In such a circumstance, a pilot may relegate a portion of his responsibility for vigilance to the controller for seeing and avoiding other aircraft. In the case of EA 111, having come from the radar environment of the approach and after having received specific landing clearance, the pilots may have experienced a natural tendency to relax in their attempts to visually search for an aircraft between their position and the intended landing runway. In any event, the Safety Board found no evidence to indicate less than expected vigilance by the EA 1}}] flightcrew. The Safety Board thus concludes that the actions of the EA 111 flightcrew, while not optimal in terms of speed control and situational awareness during initial and final approaches, were not uncommon to airline operations and were not causal to the accident. 2.4 Role of Air Traffic Control The Safety Board also evaluated the performance of the air traffic control personnel involved in this accident. The final controller was responsible for maintaining separation of succeeding airplanes on the approach to the outer marker. The monitor controller was responsible for maintaining separation of succeeding airplanes on the approach from the outer marker to within 1 mile of the runway.

[ANALYSIS] Provision of the multiple clearances to land in retatively rapid succession may have provided the north local controller with time needed to devote attention to flights waiting for takeoff clearances from runway 26 left. However, the premature clearances also make the north local controller the only controller responsible for the spacing between successive flights by removing that responsibility from the monitor controller and it may also have reduced the vigilance of the flightcrew. As a result, appropriate spacing for completion of the landings depended entirely on the continued vigilance of the local controller and the flightcrew. In this case, the flightcrew of EA 111 probably could not have seen N44UE because of N44UE’s external lighting configuration. Also, the premature clearance for EA 111 to land removed the redundancy of flightcrew vigilance when the north local controller subsequently became distracted with 36 CO 9687’s taxi clearance. Without the clearance to land, EA 111 would have had to remind the local controller that the clearance was needed. This reminder would probably have redirected the controller’s attention to tte lack of adequate spacing between N44UE and EA 11] and may have led to a correction of the problem by denying EA 11] clearance to Jand. In the 1977 edition of the Air Traffic Control Handbook, 7110.65/, the issuance of multiple landing clearances was not allowed. Specifically, paragraph 1122, “Anticipating Separation,“ stated, “Landing clearance need not be withheld until prescribed separation exists if there is reasonable assurance it will exist when the aircraft crosses the landing threshold. However, do not clear a succeeding aircraft to land on the same runway before a preceding arriving aircraft crosses the landing threshold...." This is basically the same text that is contained in the current Handbook, 7110.65F; however, the earlier procedures went on to say, "...do not clear more than the first two aircraft to land at any one time ai include traffic information with the clearance." During March 1978, isis paragraph was changed to delete the numerical limits for clearing aircraft to land. The Safety Board believes that current ATC procedures, as they oertair to the anticipated separation of arriving aircraft, require nearly flawless human performance that makes no allowance for an error of omission or lepse of attention due to any type of distractive event. Therefore, the Safety Board believes that the procedures contained in the Air Traffic Control Handbook, 7110.65F, paragraph 3-127, "Anticipating Separation," should be amended to preclude the issuance of multiple janding clearances to aircraft outside of the final approach fix.

[ANALYSIS] Although these lapses of controller performance are cited as causal, the Safety Board also has chosen to recognize that the controllers’ performance was a direct product of FAA air traffic management institutional decisions and practices that do not allow for human performance lapses in judgement or decision making. The air traffic control procedures permitted the local controller to assume full and complete responsidility for in-trail separation of aircraft on the final approach by invoking visual separation standards. As a result, two critical problems arose: (1) the separation distance between EA 111 and N44UE was reduced from the radar requirement of 2.5 miles minimum {{inside the final approach fix) to some indeterminate 35 distance necessary for N44UE to clear the runway prior to the arrival of EA 111 over the threshold; (2) an important redundant element, (the monitor controller) was removed from the loop. In addition, the air traffic procedures allow for the issuance of multiple landing clearances, which were issued in this case to CO 9687, N44UE and EA 111 in a period of 49 seconds while all three aircraft were still on final approach. An effect of this action was to remove another redundant element in the system: a1] aircraft had their landing clearances, and therefore no further communications with the local controller were necessary. When the simple but compelling distractions caused the local controller to divert his attention away from the landing aircraft, the stage was set for this accident. It is well-documented that human performance is subject to simple lapses (errors of omission), particularly in the presence of distracting events. Thus, the designers and operators of complex systems, who implement design features and operating procedures that allow one individual to assume the full burden for safety-critical operations, like the Air Traffic Control system, must bear some of tne responsibility for those accidents attributed to the occasional lapse in the performance of a single individual. Therefore, in addition to noting individual performance in the assessment of causal and contributing factors, the Safety Soard cites the broader failure of the Federal Aviation Administration to provide ATC equipment and prcecedures that adequately take into consideration those occasional tapses in human performance that must be expected. In examining the specific circumstances of this accident, the Safety Board became concerned that the current provisions in ATC procedures permit the controller to issue Yanding clearances to several aircraft in succession without assurance that the adequate separation will be maintained among those aircraft as they approach the landing threshold. Correlation of ATC communications and radar position data indicates that CO 9687’s clearance to land was given when the airplane was about 1 mile from the threshold of runway 26 right and the landing clearances to N44UE and EA 111 were given when the former was about 1/2 mile outside the final approach fix (FAF) and when the Jatter was about 3/4 mile outside of the FAF, distances of about 4.5 miles and about 6 miles, respectively, from the threshold.

[ANALYSIS] Therefore, the possibility exists that the pilot may have recognized shortly before touchdown that he was over the taxiway rather than the runway and that he may have accepted the situation because of a strongly perceived need to report for duty on time. The possibility also exists that he recognized his mistake at the last minute and was initiating a missed approach when the accident occurred. Lastly, he may have thought he was actually landing on the runway. Unfortunately, the evidence is insufficient to draw a conclusion on nis matter. 2.6 Airline Personnel Policies Tne Apache pilot elected to descend below DH and attempted to land even though he did not see the appropriate ground references. Since this decision was inconsistent with his experience and reputation as a capable pilot, his motivation for choosing this course of action was examined. After the unsuecessful first approach, the pilot could have diverted to an alternate airport, such as St. Petersburg. However, since he was commuting to work, diverting to an alternate airport would have caused him to be late for h’s scheduled reporting time. Therefore, the Safety Board examined those Eastern persc.unel policles that would have applied to such a situation and that might have influenced the pilot's decision. The Apache pilot was probably aware that the Eastern Airlines Flight Operations Manuai discouraged pilots from flying thei» personal aireraft within 12 hotrs of a scheduled Eastern flight. The chief pilot stated that the pilot may have thought that he would be disciplined for delaying his scheduled trip departure, since he had previously received a letter recommending that he "take corrective measures tc prevent a recurrence" of a delay which took place over 5 years before the accident. This delay was pete: SpE rset cn ele REN ME yee Tae ee tae twee wer fe ee ove Pe Regma TN NETS me ie OTL RAN Ra INT define gt Ne EAA OIE TEN TDA, la OM VCE TE he by OS! S Reta ey: ce? OES ee NRA US HE sone atittenar es term i OH Bt heb Bo BER ABN eat RN ST Sota deseribed as a "serious failure to fulfill your job responsibility." Although the pilot had a perfect attendance record for those 5 years, he may not have been aware of the 3-year limitation on disciplinary action for a second trip deley. Consequently, the pressure to suecessfully land at the Tampa airport, rather than to divert to an alternate airport in order to report for duty on time at Eastern probably was largely self-generated on the part of the Apache pilot. 2.7 Regulations Governing Instrument Approaches When 14 CFR Part 91 was amended in 1981, the criteria for continuing an instrument approach below DH were clarified. The revised rule specified the visual references which allow descent below DH, and it required that those visual references be "distinetly visible and identifiable." The Safety Board believes tiat this accident clearly illustrates the need to reconsider this “look-see" provision of 14 CFR Part 91.

[ANALYSIS] The pilot of N2185P was monitoring departure control (Gibbs Radar) when ABEX 110 called "Missed Approach" at 1138:31, and he was monitoring the local control frequency at 1142:27, when Cessna N66i3D wes given missed approach instructicus. By the time the pilet of the Apache executed his first approach, the touchdown RVR nad dropped to 600 feet. It is highly unlikely that he was able to see any of th: visual references required by 14 CFR 91.116 on that approach or on the subsequent approach, when the touchdown RVR was again reported to be 610 feet. The sun had risen about 20 minutes before the accident occurred. Daylight conditions prevailed, making the approach lights and sequenced flashers more difficult to see than they would have been when the other two aircraft missed their approaches. 2.3 Air Traffic Control and Approach Facilities With the exception of the failure of the local controller to issue a landing clearance during the Apache's first approach, all air traffic control procedures were correctly observed. On the Apache's first approach, a change of local controllers resulted in the failure to issue a landing clearance. The failure occurred because the two controllers used different techniques for issuing landing clearances. The breakdown in coordination between the controllers did not contribute to the events that led to the accident, but it is of concern to the Safety Board. It illustrates the need for careful coordination whenever one controller turns ever his responsibility to another controtler. The components of the ILS system were ground checked and flight checked after the accident, and they were found to be performing within the specified parameters. The alarms on the ILS components were operational and did not indicate any malfunctions on the morning of the accident. Other pilots who executed the runway 36L ILS approach that morning did not experience any difficulties with the approach facilities. The radar data recorded by Tampa Approach Control indicates that the two approaches executed by the pilot of the Apache were stable and precise. He tracked the centerline of the localizer with only minor deviations, and he applied prompt and proper corrections. On both approaches, the only significant deviations occurred at a distance from the end of the runway which corresponds to the approximate point at which the DH was reached when on the glide slope, about 0.4 mile from the runway threshold. ‘This + RRR ANE ENE RAS NR) MINE XA PHP ORNATE ts NMSA AN, SIRENS HES BARDEN NTN TENCE ETRY ERT ST NSN A Reg A AP AEN TER LEONE RAPHE NEA SEES TS SPI ~21- would be the point at which the pilot would have been jooking outside for the visual references associated with the runway and approach lights, and his attention would have heen diverted from his flight and navigation instruments. Therefore, the stability of the depicted approuches, the nuvigational aid flight test results, wad the experiences of the other pilots who flew the approach confirm that tne ground navigation aids and airborne navigation equipment were functioning properly. The approach lights were operating normally, and no alarms were noted.

[ANALYSIS] Therefore, the stability of the depicted approuches, the nuvigational aid flight test results, wad the experiences of the other pilots who flew the approach confirm that tne ground navigation aids and airborne navigation equipment were functioning properly. The approach lights were operating normally, and no alarms were noted. The alarm for the sequenced flashers activated before the accident, but a subsequent inspection revealed that only two lights were inoperative, rather than the three or more required for normal alarm activation. Investigators were unable to determine the reason for this false activation of the alarm. These two lights did not have any significant effect on the visibility of the approach light system. The other pilots executing the approach that morning saw portions of the approach lights and sequenced flashers, confirmirg their proper operation. Although the tower eab was equipped with radar, it was not designed or intended to be used for monitoring the precise alignment of aireraft with the runway during the final moments of an instrument approach. The relatively small distance between the runway and paraliel taxiway precluded any clear definition of aircraft alignment on the radar display. In summary, the Safety Board coneludes that the ILS components, the approach lighis, the airborne navigation equipment, and the actions of ATC personnel were not relevant to the cause of this accident. 2.4 Pilot's Decision to Fly The sequence of events leading to the accident began with the pilot's decision to fly, rather than drive, to Tampa. The weather briefing that he received on the night before the accident hidieated the probability of fog in Tampa on the next morning. Although the forecast did not indicate that the visibility would be below landing minimums, the Safety Board believes that a prudent pilot would have seriously considered driving to Tampa, rather than flying, under those eireumstances. If the pilot had planned for the possibility of fog by arising earlier and calling for a revised weather forecast, he would have had the option of flying if the weather was suitable, or driving if it was not. Br waiting until he was airborne before obtaining the current Tampa weather, he deprived himself of the option to drive. 2.5 Conduct of the Approaches The only information available concerning the Apache pilot's execution of the two ILS approaches is found in the radar cata recorded by Tampa Approach Control. Although the airplane was equipped with a transponder, it did not lave an altitude reporting capability or an encoding altimeter. Merefore, there is no data available to determine the altitudes flown during those approaches. The course data indicate that both approaches were flown with a high degree of precision. There was normal bracketing of the localizer centerline, bt no significant deviations were evident. As the two approaches _rogressed toward the middle marker, located 0.5 nautical mile from the end of the runway, the slight deviations decreased. Just inside the middle marker, however, at a point where the airplane should have been arriving at DH, slight excursions from the localizer were noted.

[FINDINGS] 3. CONCLUSIONS Findings 1. The flighterew of each airplane was properly qualified and certificated in accordance with current Federal regulations. Qe The Apache pilot was familiar with and had current experience in the operation of that airplane. There was 10 evidence of any preaccident failure or malfunction in the The ILS and approach light systems for runway 361 at Tampa International Airport were operating properly at the time of the accident. fee ee cen dy meme pm) ROI SF Hie OOO OORT OG A tT eI ray pain wees TA pfs creme me Ye FRR Nyaa ‘ a ¢ sete ta Ay FS eee ~25- ‘The Apache pilot was told of the potential for fog at Tampa when he was briefed on the r “ht before the accident. The Apache pilot did not call for 4 eather vriefing before taking off on the morning of the accident. The presence of fog was confirmed after the Apache pilot was airborne. The two pilots who executed approaches to Tampa shortly before the accident were unable to see any of the required visual references related to runway 36L when they reached decision height. The captain of the Pan American B-727 observed a maximum visibility of about 500 feet between the taxiway and runway just before the accident. The aecicent occurred in daylight conditions, which diminished the effectiveness of the approach lights. The failure of the local controller to issue a landing clearance during the Apache's first approach did not contribute to the accident. The Apache pilot flew Doth approaches with precision until Ne passed the middie marker, after which slight deviations from the localizer occurred. During the second approach, the Apache pilot continued his descent below the published DH and landed his airplane on the taxiway. The Apache pilot probably was highly niotivated to land his airplane at Tampa airport despite the less than minimum visibility required because of his perceived need to report for work on time. The extent to which recent changes in personnel policies and working conditions influenced the pilot's decision to continue the approach below DH cannot be determined. 3.2 Probable Cause The National Transportation Safety Board determines that the probable ccuse of this accident was the decision of the pilot of the Apache airplane to continue a precision instrument aoproach below the published decision height when the required visual references were riot distinctly visible and identifiable. Contriduting to the accident was the pilot's failure to obtain a predeparture weather briefing before choosing © means to travel to his destination.

[ANALYSIS] There was normal bracketing of the localizer centerline, bt no significant deviations were evident. As the two approaches _rogressed toward the middle marker, located 0.5 nautical mile from the end of the runway, the slight deviations decreased. Just inside the middle marker, however, at a point where the airplane should have been arriving at DH, slight excursions from the localizer were noted. On the first approach, this deviation was to the left (west) of course, and on the second approach, the airplane moved to the right (east) and lined up with taxiway W. At this point, the pilot would probably have been looking for the visiial ~22- references that would have allowed him to continue the approach below DH. If those references were not immediately recognized, he should have initiated a missed approach. Title 14 CFR 91.116 requires that at least one of the specified visual references related to the landing runway is distinctly visible and identifiable at DH and that it remains visible for the remainder of the approach. The requirement is intended to provide protection against the pilot's disorientation during the most difficult portion of the approach--the transition from instrument references to visual references that are adequate to effect a landing. The required minimum visibility for an approach procedure is based on two factors: the accuracy of the available :adio navigation aids and the lighting systems installed on a given runway. The combination of DH and minimum visibility, generally referred to as landing minimums, is designed to allow a pilot to descend on instruments to a point at which visual references will allow the pilot a normal descent to lending on the intended runway. The fact that the pilot of the Fiper Apache landed on the parallel taxiway indicates that he descended below DH without having any of the required visual references related to runway 36L "distinetly visible and identifiable" as specified in 14 CFR 91.116(e\3). Therefore, the Safety Board concludes that during his second approach, the Apache pilot intentionally descended below DH in an attempt to visually identify the runway. He inadvertently flew to the right of the ILS localizer during the time when nis attention was divided between the cockpit instruments and the search for outside visual references. Whether he realized before touchdown that he was aligned with the parallel taxiway or thought he was landing on the runway cannot be determined conclusively from the existing evidence. However, the differences in width, surface color, and lighting betwaen the taxiway and the runway should have been apparent to the pilot at some point during the flare and before touchdown, particularly since tie pilot was very familiar with the Tampa airport. Therefore, the possibility exists that the pilot may have recognized shortly before touchdown that he was over the taxiway rather than the runway and that he may have accepted the situation because of a strongly perceived need to report for duty on time. The possibility also exists that he recognized his mistake at the last minute and was initiating a missed approach when the accident occurred. Lastly, he may have thought he was actually landing on the runway.

[ANALYSIS] In reduced visibility weather conditions, Promech pilots typically fly southwest over Ella Lake and proceed beyond the south end of the lake before turning west. This route allows the flights to avoid the high terrain located west of the lake, which includes two similarly shaped mountains. (See figure 16.) Figure 16. The typical flight route (white) passes two similarly shaped mountains before turning west. For comparison, the path flown by the accident pilot is shown in green. Although the accident pilot climbed the airplane to an altitude that would have provided safe terrain clearance along the typical route, he deviated from that route and turned the airplane west after it passed only the first of the two mountains. The pilot’s route deviation placed the airplane on a collision course with the 1,900-ft second mountain, which it struck at an elevation of about 1,600 ft msl. (See figure 17.) NTSB Aircraft Accident Report 47 Figure 17. The path flown by the accident pilot (green), viewed from the approximate altitude of the accident airplane, is shown turning to the west (right) after passing the first mountain. Note: A translucent mask is imposed on the image to provide a notional representation of reduced visibility weather conditions. The vantage point of the image distorts the flight route depiction in the foreground (where the segment of the route originates in the bottom center of the image); there is no steep vertical climb in the flight route at the location shown in the foreground. Research suggests that, when navigating, people find their way through large-scale environments by learning to identify key landmarks (route knowledge) and developing map-like internal representations (cognitive maps) of the area. When traveling, people use the landmarks to update their position on their internal cognitive map and decide when and where to turn (Golledge 2010). However, a number of factors can increase the likelihood of navigational errors. Navigational errors are more likely if landmarks are difficult to discern or if cognitive maps are inaccurate (Golledge 2010). Landmarks are more difficult to discern when visibility of fine detail is degraded or when landmarks are viewed from an unfamiliar vantage point (Cuqlock-Knopp and Whitaker 1993; Schreiber et al. 1998). Several factors were present during the accident flight that increased the likelihood that the pilot would make a navigational error.35 Visibility was degraded over the south end of Ella Lake; thus, the pilot may not have been able to clearly distinguish some of the landmarks by which he typically navigated. In addition, the accident pilot was flying the route at a relatively low altitude because of the poor weather conditions, which resulted in his viewing terrain from a lower-than-normal vantage point. (The accident pilot had transited the area earlier that day about 2,000 ft msl, and other Promech pilots had flown about 2,500 to 3,000 ft msl.) Further, the accident 35 In addition, some pilots reported that it could be difficult to see directly ahead in the DHC-3 because the turbine engine conversion resulted in less efficient clearance of rain from the windscreen. However, investigators were unable to independently evaluate these claims.

[ANALYSIS] However, investigators were unable to independently evaluate these claims. NTSB Aircraft Accident Report 48 pilot’s relatively short time flying in the local area suggests that his cognitive map of the terrain between Rudyerd Bay and Ketchikan may not have been as detailed or accurate compared to those of other pilots who had more local flying experience. Considering all these factors, it is possible that the accident pilot mistook the first mountain for the second one and turned the airplane west too early; such a mistake, in which a pilot does not recognize or maintain the desired position relative to the external ground and airspace environment is an example of geographic disorientation (Antuñano, Mohler, and Gosbee 1989). The fact that the pilot subsequently continued the flight straight and level at an altitude that would have been sufficient to clear terrain on the typical route (but was insufficient to clear the high terrain he was flying toward) further supports the possibility that the pilot was geographically disoriented. Thus, the NTSB concludes that the pilot’s continued flight in low-visibility conditions at a lower-than-normal vantage point in an area in which he lacked extensive flying experience reduced his ability to visually identify landmarks and resulted in a navigational error due to geographic disorientation. 2.2.3 CFIT According to recorded flight data, the airplane’s final turn to the west occurred about half-standard rate; then the airplane continued on a nearly straight and level flightpath toward the higher terrain for about 30 seconds. In the final 2 seconds of the flight, the airplane pitched up rapidly, experiencing a vertical acceleration of about +2 Gs before colliding with terrain. The timing of the pilot’s aggressive pitch-up maneuver strongly supports the scenario that the pilot continued the flight into near-zero visibility conditions and, as soon as he realized that the flight was on a collision course with the terrain, he pulled aggressively on the elevator flight controls in an ineffective attempt to avoid the collision. Effective CFIT avoidance requires assessment of weather conditions and appropriate aeronautical decision-making in response to cues associated with deteriorating weather. The pilot missed opportunities to prevent the accident both in his decision to take the short route despite the presence of low clouds and obscured terrain, to turn toward an area of the lake where visibility was further reduced, and to continue the flight into deteriorating conditions rather than escape them.

[CONCLUSIONS > FINDINGS] NTSB Aircraft Accident Report 68 20. The establishment of a more conservative set of weather minimums that are tailored to the types of air tour operations that occur in Ketchikan and are applied to all area air tour operators would help balance competing goals of production and safety and remove the incentive of individual operators to adopt the lowest possible weather minimums to stay competitive. 21. Automatic dependent surveillance-broadcast information from a focused operational area could serve as a basis for the Federal Aviation Administration to promote open, nonpunitive discussions of safety issues reflected in objective data to increase awareness of operational hazards. 22. A robust discussion by the Cruise Lines International Association’s safety committee with its members about the scheduling pressures associated with air tour shore excursions identified in this investigation would benefit the entire cruise industry. 23. A flight recorder system that captured images of the pilot’s forward view would have benefitted this accident investigation and provided potentially valuable information for air tour operator training programs. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was (1) the pilot’s decision to continue visual flight into an area of instrument meteorological conditions, which resulted in his geographic disorientation and controlled flight into terrain; and (2) Promech’s company culture, which tacitly endorsed flying in hazardous weather and failed to manage the risks associated with the competitive pressures affecting Ketchikan-area air tour operators; its lack of a formal safety program; and its inadequate operational control of flight releases. NTSB Aircraft Accident Report 69

[ANALYSIS] During the accident tour, the accident pilot chose to follow another Promech pilot on the short route, even though the route presented an increased risk of an inadvertent encounter with IMC and increased the risk of CFIT. The pilot’s decisions to fly the short route and to continue flying once he encountered deteriorating weather were likely influenced by several operational factors. 2.3.1 Company Tour Schedules and Practices of Other Pilots About 90% of Promech’s business consisted of air tours, and the company competed with other area operators for cruise line contracts; thus, the company was likely motivated to provide service that would be highly rated by the cruise line passengers, including returning the passengers to the ship on time. In addition, tour operators that failed to return passengers in time for a ship’s departure were responsible for the accommodations and travel expenses associated with delivering those passengers to the ship’s next port of call, providing further motivation for punctual returns. The accident tour departed late, and Promech’s president/CEO had reminded the pilots to be mindful of the cruise ship boarding time. Three of the four Promech flights that departed together on the return leg for Ketchikan followed the short route; the only exception was one pilot (the second pilot) whose personal risk tolerance led him to conclude that the weather was too hazardous. (This pilot wanted to avoid reversing course as he had done earlier in the day [during the return leg on the first tour] because of weather over Ella Lake.) Thus, schedule pressures likely motivated the pilots in the accident tour group to return to Ketchikan using the most direct route, if possible. In addition, the two pilots flying their airplanes (one from Promech and one from another operator) a few miles ahead of the accident pilot chose to fly the short route over Ella Lake and broadcast their intentions by radio to other pilots on the frequency. (The pilot who chose to take the long route did not announce his decision.) The Promech pilot who took the short route ahead of the accident pilot was highly experienced and a close friend of the accident pilot; colleagues said the accident pilot admired and tried to emulate him. The other operator’s pilot was also highly experienced with flying in Alaska.

[ANALYSIS] The NTSB concludes that the airplane had no preimpact anomalies that would have precluded its normal operation. The crew and passenger seat structures were severely damaged and fragmented due to the high loads sustained during the terrain collision, and the floor and sidewall structures of the 34 The pilot’s autopsy identified mild focal narrowing in the pilot’s three main coronary arteries, which was not significant and was unlikely to cause symptoms. Although the autopsy also identified focal intraventricular septal fibrosis (scar tissue), a condition that may increase the risk of irregular heart rhythm and possible incapacitation (Morita et al. 2014), the circumstances of the accident, including evidence that the pilot suddenly and aggressively pulled on the elevator flight controls in the final seconds of the flight (described in section 2.2.3), are inconsistent with incapacitation. NTSB Aircraft Accident Report 45 airplane were damaged to the extent that they could no longer anchor the seats. Thus, the NTSB concludes that, based on the damage and high loads sustained during the impact sequence, the accident was not survivable. 2.2 Accident Sequence The accident flight departed at 1207 and was the third of four Promech-operated float-equipped airplanes that departed within minutes of each other from Rudyerd Bay. The accident flight and the two flights that preceded it were returning cruise ship passengers to Ketchikan for a 1230 “all aboard” time; the fourth flight had no passengers but was repositioning to Ketchikan for a tour scheduled at 1230. All of the flights were running behind schedule from the intended 1145 departure time. The accident pilot and two others chose the short route between Rudyerd Bay and Ketchikan (which takes about 25 minutes), while one pilot (the second Promech flight to depart) chose the long route (which takes about 30 minutes). (Section 2.3 provides more information about pilot decision-making regarding route choice.) After landing, the airplane must be taxied to and secured at a dock for the passengers to disembark and board a bus for a 5- to 7-minute ride to the cruise ship. 2.2.1 Weather Conditions About the time of the accident, multiple weather observation stations in the area (each located about 20 to 30 miles from the accident site) were reporting MVFR conditions (and one was reporting IFR conditions) with visibility restricted in rain and mist. Satellite imagery showed an extensive area of low clouds over the region, including a broken-to-overcast layer with some vertical development consistent with some clouds supporting rain showers. Images from FAA weather cameras located in Misty Fjords (10 miles east of the accident site) and Ketchikan (23 miles southwest of the accident site) showed that the higher ridges were obscured by clouds and that, at times, poor visibility and/or low cloud layers existed in some areas. The NWS area forecast valid for the accident area for the period before and during the accident flight included an AIRMET advisory for mountain obscuration with MVFR and temporary IFR conditions due to light rain, mist, and low ceilings.

[FINDINGS] He also was medically qualified. The weather was clear at the vollision altitude. The «airplanes were opereting under visual flight rules. N3827C was being rader vested by the New York Terminal Radar Approach Control for a practice ILS (instrumnont Janding system) approach to runway 6 at Teterboro airport. The pilot of N96402 did not have radio contact with an afr traffic facility. The cockpit visibility study indicated that during the 45-se:ond period before the collision, the detection of N96402 was restricted by the windshield centerpcst in the vision envelope of N3827C's pilot, and that during the 15second period before the collision, the image of N3827C was unrestricted in the forward vision exvelope of the pilot of N96402; however, in the prior 30second period, N$827C was in only the monocular vision of N$6402's pilot. During the last 30 to 45 seconds before collision, neither pilot had a totally wwbdstructed view of the other airplane, at least until the targat s’ze filled the winishield at some time between 15 seconds and collision. During the 107 seconds before the collision, the passenger in N3827C had the iznage of N96402 in full view near his sero wye reference. The clicumstances of this accident Involve problems associated with the Umitations of human vision and the inherent difficultles of perceiving, recognizing, and effectivaly avoiding a collision with another eirplane. If both airplanes had been equipped with altitude encoder devices, the controller would have been better able to recognize the potential conflict of the two airplanes. 1$. The controller could have axd should have observed the potential traffic conflict and issued an appropriate advisory. 14. The Safety Board determined that the collision occurred in the controlled airspace of the New York Terminal Control Ares. Probebla Cimuse The National Transportation Safety Board determines that the probable cause of this accident was the failure of the pilots to exercise adequate vigilance to detect and avold eeh other. The failure of the pliots may have been due to the limitations of hum n vision and the Inherent difficulties of perceiving, recognizing, and effectively avoiding a collision. Contributing to the accident was the failure of the pilot of N96402 either to keeg: clear of the New York Terminal Control Area or to avail himself of the traffic advisory capability of the New York Terminal Radar Approach Control Also contributing to the accident was the failure of the controller to observe the potential conflict and to adequately convey traffic information to N3827C.

[ANALYSIS] Normally, an instrument-ratod pilot could be expected to maintain the last assigned altitude until directed by the controller to descend. In this case, the controller would not have deseended the airpiane farther until the alrplene wes established inbound on the ILS localizes and wan past the Dandy Intersection. The pilot haa this information on an instrument approach chart in his pousession. At the time of the collision, N3827C had not reached the desent points. Thus, as the preponderence of evidence supports the conchision that the collision occurred in the controlled airspace of the TCA, the Safety Board determines that the accident occurred within the boundaries of the TCA at an nititude of about 3,000 feat. The collision damage to N3827C was consistent with witness accounts that the front of N96402 collided with the side of N3827C's rear fuselage and sheared off the latter's pmpennage. Paint transfer marks and inward crush damage on N3827C indicated that it was hit from the right (see appendix C). Propeller slicing across the rear fuselage and the left stabilizer of N3827C revealed that N96402's exgine went across N2847C's tall section nt a S$-degree angle. Based upon the ATC radar tracking data, the relative approach angle wes 120 dogrees, and the 55-deyree propeller marks indicate that N96402's heading ‘was turned significantly to the right when it struck N3827C's fuselage. This direction wshange $s believed to have been due to the evasive maneuver by N96402's pilot and the slewing of N96402 efter an initial collision between the right wings of the airplanes. In view of the favorable weather conditions and the angles of approach, the Safety Board could not determine why both pilots did not see each other. The Board recognizes that elthough both pilots may have been scanning regularly for other traffic, they may have been distracted at a critical time by chart reading or cockpit functions that interruoted their outside scan pattern. Additionally, the pilot of N3827C may have been overconfident that the TRACON controller was protecting his airspace because his airplane had been redar identified, his altitude had been acknowledged, and he was flying in positive controlled airspace. Although the position of the sun at the time of the accident was low on the horizon and slightly to the right of the track of N3827C, the Safety Doerd believes that because of the high overcast, the glare of the sun would not have reduced the visuel range normally available to the occupants of N3827C. The sun would have been behind the pilot of N96402, and it would not have affected his ability to see. There was a very limited period of time (107 seconds) for target detection.

[ANALYSIS] These targets were not evidenced on the ATC radar data recorde? during the perlod, possibly because they were not transponder equipped. N96402 would have been between the 1 o'clock and 2 o'clock positions, at a range of about 7 miles, The Safety Board could not determine whether it was among the targets perceived by the controller. After Ns827C completed its southerly turn and was established on a collision course with N964U2, the controller did not identify the N96402 target as conflicthg traffic to N3827C's pilot and, therefore, did not issue a further radar traffic advisory. The Safety Board notes that the recorded ATC radar data provide evidence that the target cf N96402 was present on the controller's display. The Board believes that the controller could have and sould nave observed the potential conflict and issued an appropriate advisory. Since the controller states that he was continuously monitoring the radarscope, the Board cannot determine the reason the controller did not recognize the potential conflict along the edge of the TCA. The Board conchided that the collision potentis] was evident before either airplane crossed the horizontal TCA boundary. -5.Y Seg a tr tacnsastttals deen weceneneasten ete ens etn i ree erence se ee we ee ocean wpanben maces te mensae er we ew coe rmeine stam Gh mime peel ciate ct COETS Th alineinee NealieetiRan —apreniemmertin Atiitinindtimcmmns cen em «wie fee tn -l1- Although there was no definitive evidence to confirm the collision altitude, there wore several circumstances relating to the altitude question. The New York TRACON received th last encoded transponder altitude of N3827C at 1611:39, 2 minutes 20 seconds before the collision. At 1612:04, 1 minute 55 seconds before the collision, the pilot of N3$837C told the controller that he was at 2,000 feet, and the controller then told the pilot to maintain that altitucle, and, in the same transmission, called traffic. The pilot acknowledged by saying, "okay roger sir, that's Caldwell Airport." Because the pilot did not verbally repeat the altitude, it {{2 not known if he received that part of the controller's transmission. However, the pilot's reply iaay have been o singia statement, which responded only to the traffic advisory, or his statement may have been in two parts: (1) “Okay roger, air,...", which could have been an acknowledgement of the altitude flearance and {{2}} "... that's Caldwell airport", which could have been in response to the aivisory of the mimerous targets at the plilot’s 12 o'clock position. Normally, an instrument-ratod pilot could be expected to maintain the last assigned altitude until directed by the controller to descend.

[ANALYSIS] The Board considers this to be as important as the familiar subjects of map reading, weather symbology, and pilotaege. G7 Alreraft Accident Report--"North Central Airlines, lie., Allison Convair 340/440 (CV80), N92858, and Air Wiscontin, inc., DHC-6, N4043B, near Appleton, Wisconsin, June 29, 1072" (NTSB-AAR-73-0), es eh aed ae ee rie te re Be Fal Merman ut RRR Rall OOF Wh deere WtIes tee The system of providirg separatica is not error-proof, nor In all predaebllity will it errte be. Conflicting traffic, particularly near the boundar'ss of a FCA, mey be a threat diiectable only by pilots, end S:en only If they ere looking for it. Thete may be aie cuinraom denominator to all midair eollindons, and that factor night be described as pilot coiplecency particularly when an alrplene is under positive control. The Safcty Bowd euohecies as an essential part of a collision avoidance program that separation can be nisintained most effectively by pilots who recognize that outside scanning must be an argrersive procedure. Target recognition is a difficult task, aad pilots must lean to train themselves to use head and body raovamente as well as eye movements in a planned scerning pattern to overcome the linitations on target detection in order to de able to teke timely evasive action. CONCLUSIONS Findings 1. The airplanes were certificated, equipped, and n.aintained in accordance with Federal regulations and approved procedures. 2. About 2 .ainutes before the coliision, the New York Terminal Radar Approach Control radar ceased recording transmissions from the altitude encoder of N3827C. N&6402 was .ot equipped with an altitude encoder. The pilots were certificated properly. There was no evidence of preexisting medical or physiological problems that might have affectad their performance. The radar controller was qualified as a full performance level controller. He also was medically qualified. The weather was clear at the vollision altitude. The «airplanes were opereting under visual flight rules. N3827C was being rader vested by the New York Terminal Radar Approach Control for a practice ILS (instrumnont Janding system) approach to runway 6 at Teterboro airport. The pilot of N96402 did not have radio contact with an afr traffic facility.

[ANALYSIS] The pilot of N96402 was operating under VFR throughout the flight. There was no evidence that he had besn in. «dio contact with any flight service stetiun or tower facility along his route of flight. How. ver, his airplane's transponder was transmitting the YPR identifice code, whish could have identified his ulrplene as V?R traffic on rader. The contzolier said that there were numerous radar targets in the area, but that he could nct rocall the movements of any of the VFR targets. Although the target of N998402 would have been on the display, the controller apparently never recognized or perceived it as potential traffic fo: the aircraft that he wes working. 196402's track, as plotted cn the pilot's New York Aren chart, touched the TCA boundary where the 3,000:-foot floor level lowers to 1,800 feat. The proximity of the pilot's rnarked routs to the TCA boundary provided no margin for Gead reckoning navigational error to ensure that the airplane would remain outside of the TCA airspace, unlees the pilot flew through the eroa at an altitude well below 1,800 feet. The point of collision was about 1 mile east of the plotted course. The Sc.ety Boerd believes that N96402 was operating tou closely to the TCA airspace, especially considering the fact that the pilot elected not to communicate with the facility having ‘urisdiction over that airs.sce. The periphery of a TCA should be viewed with great caution as it is here that nak. Na nl A A oe Al Mi nb A RO ES EN OT ST RN Re - ARR A OREN ia nef manlemiit A oth AAI AN OTE’ Malta there is the greatest likelihood of intermingiing of controlled and uncontrolled traffic. At 1611:42, N3827C was issued a heading of 170 degrees. The turn was completed at some point between the 1612:07 radar "hit" and the 1612112 radar hit. At 1612:07, the controller transmitted "okay maintain two thousend ah numerous targets in your twelve o'clock position one showing a thousand fest altitude unverified the other altitudes unknown." N3827C's pilot then transmitted "okay roger sir that's Caldwell Airport." At this point, Celiwell Airport would have been in N3827C's 9 o'clock or 10 o'clock positions, wot the 12 o'clock position. The Sarety Board believes that the pilot was not looking at the traffic called by the controller. This circumstance shoukl have been recognized by the controller, and he should have clarified the position of the 12 o'clock traffic to the pilot. These targets were not evidenced on the ATC radar data recorde? during the perlod, possibly because they were not transponder equipped. N96402 would have been between the 1 o'clock and 2 o'clock positions, at a range of about 7 miles, The Safety Board could not determine whether it was among the targets perceived by the controller.

[ANALYSIS] He was working several airplanes and he had admitted to being busy. At 1718:39, when the PA28 piiot reported ".. .one mile to the west" he asked the pilot to ", . weport clear of the airport traffic area to the west" when he should have said east. The Safety Board believes that the local controller inadvertently mistated the location of the PA28. Since he was not corrected by the PA28 pilot in his acknowledgment, the controtier had no reason to believe thereafter that the mistatement was incorrect. Therefore, he repeated the mistake on two subsequent instances. Furthermore, he did not provide the DA5O flighterew with the type identification for N1959T nor for the PA28, as preseribed by ATC procedure. This information could have been helpful to the DA50 flighterew in identifying the traffic. Hcwever, the Safety Board believes that the most significant error on the part of the local controller was not reporting that the additional traffic was an eastbound overflight. This type of human error is indicative of a pronounced workload condition. Since there were several airplanes in the traffic pattern at the time of the cojlision and since it is not completely clear which airplane the flighterew considered to be the "westbound" reported traffic, the Safety Board identified and analyzed three aceldent hypotheses that could have evolved from the conditions and circumstances presented to the DA50 crew and the local controller. These hypotheses are based on the belief that the DA5O flighterew were not aware that the local controller had sequenced N1959T between them and the Twin Cessna. First Hypothesis.—The first hypothesis assumes the DA5O flighterew identified N1959T as being the number two airplane to land and they would be number three. In this case, they would have had to choose between the PA28 and N72BG as being the additional traffie reported to be "westbound." At 1719:45, the captain then was referring to N1959T when he sald, "Better slow it up Alan, we're following that guy." This airplane, as well as N72BG, was a light twin engine airplane. Furthermore, because it was closer in his 12 to 1 o'clock position at 1,500 feet and it was a light single engine airplane, it is postulated that the eaptain was refercing to the PA28 as the "westbound" traffic. At 1720:18, when he commented, "You're eat'em up" he then was referring to the FA28 indicating that they were rapidly converging. This comment prompted the right turn made immediately by the first officer for separation. Also, the flightcrew would not have been surprised when they passed the PA28 and they would have assumed that it would no longer be a factor. Twenty-one seconds later, at 1720:39, the captain referred to another aircraft down low and the CVR recorded an unintelligible response from the first officer.

[ANALYSIS] Airplanes were frequently given permission by the control tower to transit the airport traffic area in order to reach the Hudson River. The pilot's decision to transit that night at 1,500 feet was risky because of the overhead approach in operation at TEB and the potential conflict with turbojet traffic. Even though he had the VOR/DME~A frequency tuned in, he may not have known the approach was in use at the time. It is suspected that he chose 1,500 feet in order to remain below the floor of the TCA and above the small alreraft traffic pattern. His decision not to contaet the TRACON and obtain radar service was evidently governed by his opinion of radar service in the area and the time he had available to make the flight with the expectation of being given a circuitous route to the Hudson River had he done so. The fact that he made a mistake In his position report west of the airport would have been significant only if the local controller had known about the DASO earlier. Furthermore, had the pilot corrected the local controller's mistake when he asked the PA28 to "report clear of the airport traffic area to the west," the pilot may have stopped the controller's "mindset." Besides, the pilot should have been concerned about the controller's knowledge of his flightpath in order to obtain timely and accurate traffic advisories. However, it is possible that radio frequency congestion might nave precluded the PA28 pilot from correcting the local controller. Since the DA50 was behind the PA28 initially, it {{s doubtful the PA28 pilot ever saw the airplane. He also had the right-of-way in this case. When the local controller tried to caution the pilot about the DA5O at 1720:26, the response of the PA28 pilot was believed blocked by another radio transmission. Also, if he heard that the DA50 crew had his airplane in sight, he probably felt no need to take any action. The DA60 passed the PA28 on its right and made a right turn away from the alrplane. Therefore, until moments before the collision, the pilot would have had difficulty sighting the airplane, particularly with a passenger in the right seat who would have obscured some of his vision. Five seconds before the collision, the pilot was telling the controller that he was clear of the area to the east as requested. Therefore, it is believed that he did not see the airplane until {{t was too late to avoid the colliston. WRSR EM LEE 4 RAG OTR SAM MAL HTN Haag OM NRT OEE ome Me oe In view of the traffic situation and the znvironmental and physiological limitations that were present, it is evident that the pilots would have had difficulty detecting the traffic and discerning their flightpaths. Under these conditions it takes longer for a pilot to detect conflicting traffic, make a decision, and take evasive action. The FAA's estimate of 12.5 seconds for such action to take place does not take Into gecount the extended time required at night. When the captain said, "Hey, watch out.

[ANALYSIS] This apparent assurnption would have influenced their expectation of the traffic sequence in the pattern. Furthermore, the Safety Board believes that the excessive speed and the flighterew's preoccupation with the PA28 resulted in their losing awareness of the position of the DA50 with respect to the airport and the normal limits of the traffic pattern. Had the captain recognized that his airplane was well outside the limits of the traffic pattern when the first officer was obtaining separation from the PA28, the captain should have known that the PA28 was not in the .traffie pattern and, therefore, was not the airplane that the local controller had identified as number two to land. me i nar mT ON NE NES EB Si A a A RN NSE NLR oc Rey EEN ITD . “1 ‘i : a * a were Ae fran crgpeg te gen Consequently, the Safety Board concludes that the flighterew lost "situational awareness" because of a preoccupation in flying the airplane, they had a preconceived idea that they woutd be following a Twin Cessna in the traffie pattern, they received misleading and inadequate information from the local controller, they were misled by the position of the airplanes they observed in the traffic pattern, and they failed to relate the position of their airplane to the airport. These factors led to their misidentification of the PA28 as the airplane ahead of them in the landing sequence and led to the {{improper assumption that the PA28 would make a left turn onto a downwind leg for runway 19 both of which established the condition for the collision course with the PA28. Therefore, the Safety Board conicudes that the flightcrew's loss of situational awareness was a significant factor In the accident. CLIFO is only a mandatory reporting point on the instrument approach to TEB when an aircraft is not in radar contact with ATC. Since radar service was not terininated by the departure controller, the flighterew was not required to make 8 position report at CLIFO based on radar status alone. The departure controller instructed the DA5O to contact the tower at CLIFO. However, the radio frequency congestion prevented a timely position report to the local controller. The clearing turn to the right was the proper maneuver in an overtaking situation. It also would have been the proper turn to make In a converging situation. The CVR record otherwise disclosed that the flighterew was active in sighting traffic and followed their preseribed procedures. The manner in which the PA28's flight was conducted was vonsidered routine. Airplanes were frequently given permission by the control tower to transit the airport traffic area in order to reach the Hudson River. The pilot's decision to transit that night at 1,500 feet was risky because of the overhead approach in operation at TEB and the potential conflict with turbojet traffic. Even though he had the VOR/DME~A frequency tuned in, he may not have known the approach was in use at the time.

[ANALYSIS] However, N1959T was more than 20° to the left of the PA28 and therefore it would have appeared to have been the first airplane {{n the pattern for landing and with the PA28 following as the second airplane in the pattern. Furthermore, the only other airplane that was inconsistent with the flow of traffic in the pattern at that time, and within the flighterew's view was N72BG. Therefore, it is believed that the DA50 flighterew may have mistaken N72BG for the additional traffic reported to be "westbound." Having observed the PA28 for more than 20 seconds, the crew was aware of its easterly heading and would, therefore, have had no reason to suspect it was the subject of the "westbound traffic" advisory. Their response, "We're looking," is understandable, given the number of airplanes in their view. The Twin Cessna they may have thought they were following was well ahead of the other airplanes and about to turn onto the base leg at this time and was not considered e factor. ~AY~ This hypothesis takes into account the belief that the captain perceived an overtaking situation with the PA28 and is consistant with the right turn for separation. Thinking that the PA28 was the Twin Cessna they were following, they established an. interval on that airplane by making the right turn and were anticipating that the PA28 would make a left turn for the downwind leg. As 4 result, the flighterew became preoccupied in obtaining separation with the PA28 they thought was number two to land and was unaware they were 3 to 4 miles southeast of the airport and outside the traffic pattern. This hypothesis also accounts for the first officer's question, "What kind of Cessna is that?" because they thought they were following a Twin Cessna and the PA28 was the closest airplane from which an identification might have been made. It should be pointed out that the Safety Board also recognizes the possibility that the first officer could have raised the question because of the PAZ8's relatively slow speed verses that which would be anticipated from a Twin Cessna and perhaps this was the reason. However, because of the congestion on the radio frequency, the captain did not get a chance to ask the controller to resolve the ambiguity. In the Safety Board's opinion, the local controller's failure to report the additional traffie as an eastbound overflight {{ts significant because it is not likely that the flightcrew would have made the left downwind turn into traffie they were told was eastbound at the same altitude. In addition to providing the eastbound overflight information, had the DA50 crew been given the identification of N1959T, the airplane they were supposed to follow, they would have been confronted with a different situation and it would have caused them to question the presence of the PA28.

[ANALYSIS] This was considered a proper action based on preseribed ATC procedures. However, the local cortroller's failure to mention that the additional traffic was an eastbound overflight instead of "westbound" is considered to be a factor in the accident. OR ey Sr NAAN ER ONES AMY at MORAN A oy . ET ah EN whe 8 YO SR Ny yh EG So orl tt ences aly a ttn 2 AS ot TE ae oe Pt RS Se RG ilar RE ag MEN te i A Ee a ane Ln alge ata get ape tre A aac hae tations 3 a tole hot Sa TS, AZIMUTH (DEQ) TWAE 17:21:00 ~90 +20 ~-10 ) 19 cr a ae Aveta ae ales L ELEVATION WES}} AZIMUTH (DEG) 17:21:00 y N (TRUE) Compr WHAT KIND OF CESSNA IS THAT? y CAPTAIN I DON'T KNOW 7 ~— S WrIBG = «1300 % “ “a ST PAW 1700 0,0 0,5 1,0 N . ‘e \ teit In Nu. cate 1600 7 \ S \ V7¥2106 ‘ a CAPTAIN FLL ARK Him ‘“ Figure 22.——Cockpit radar positions and cockpit visibility plot of aircraft at 1721:00. TAME 17:21:22 AZIMUTH (DEQ) “30 -20 ~t6 0 wc 20 =36 ~-20 -40 G6 10 20 AZIMUTH (DEG) Rd TRUE) $ret MIS7TH — -TETERBORO TOWER CHEROKEE ONG nae 0.6 0,$ 1,6 SEVEN SEVEN HOTEL CLEAN TO Tee BAST tceTs In Nad, s PALa tro ms Sate 1400 S nN 17 :tt aw ‘ COPMOT LETS GO rML Papa" *@- - “ / “N engias WEY WATSH OUT, THI GUY'S COeMM CAPIAW GO DOWN MONT AT US Figure 23.—Cockpit radar positions and cockpit visibility plot of aireraft at 1721:22. The local controller probably reported the PA28 as "westbound" and failed to mention that it was a transiting airplane because he was busy and the radic frequency was congested. He was working several airplanes and he had admitted to being busy. At 1718:39, when the PA28 piiot reported ".. .one mile to the west" he asked the pilot to ", . weport clear of the airport traffic area to the west" when he should have said east. The Safety Board believes that the local controller inadvertently mistated the location of the PA28.

[ANALYSIS] However, it is assumed that the "westbound" report was not significant because the flighterew did not comment about head-on traffic at the same altitude, and the captain's comment, "You're eat ‘em up," implied the perception of an overtake situation. Therefore, the flightcrew discounted the two "westbound" reports made by the controller as a mistake. Although this hypothesis would account for the first officer's comment at 1721:00, "What kind of Cessna is that?" the Safety Board believes the hypothesis also does not fully account for the confusion that would have arisen with the appearance of N72BG relative to the PA28. The Safety Board believes that at least from 1720:26 to 1721:22 the DASO crew would have believed there was additional traffie because both airplanes remained In view. When the first officer began the downwind turn, the relative positions of the PA28 and N72BG again would have. made it difftcult for the flighterew to determine which airplane was number two to land. Third Hypothesis.--The Safety Board believes that a third and more likely hypothesis is supported by the available evidence. Because of the relative positions of the airplanes when the DAS5O0 flightcrew first established communications with the local controller, it assumes that the flighterew believed that the PA28 was number two to land behind N1959T and they were number three based on the captain's comment at 1719:45, At this time, the DA50 was just completing a left turn to intercept a 305°~radial of the VOR/DME approach; the PA28 was on a heading of 122° just south of the VOR. Because physical evidence indicated that the p..{{tion lights may not have been on, it could have been extremly difficult for the DA50 crew to recognize that the airplane they were behind was a small airplane, let alone a PA28. Therefore, it would be logical for the flighteraw to assume that an airplane on the same track over the airport would be the traffie they were following into the pattern, the Twin Cessna. Again, they did not know the controller had cleared N1959T between them and the Twin Cessna. To further reinfores this thought, the DAS0 crew was told at 1720:05 "Okay, plan number three following traffic tuning downwind abeam the tower...." At this time, three aircraft were within the crew's view and both N1959T and the PA28 could have been perceived as being abeam the tower. However, N1959T was more than 20° to the left of the PA28 and therefore it would have appeared to have been the first airplane {{n the pattern for landing and with the PA28 following as the second airplane in the pattern. Furthermore, the only other airplane that was inconsistent with the flow of traffic in the pattern at that time, and within the flighterew's view was N72BG.

[FINDINGS] 3.1 Findings NN ee ES Pe Fe ee OD 1, The pilots {{n both airplanes were qualified for their respective flights and were familiar with the operations at TEB. There were no apparent medical factors affecting their performance. 2, The controllers were qualified as fult performance level controllers. There were no apparent medical factors affecting their performance. 3. The pilot of the PA28 elected to transit the airport traffic area at 1,500 feet which was in potential conflict with turbojet traffic pattern. 4, The DA5O flighterew did not comply with a speed reduction to 180 knots for several minutes following the request from ATC and were 15 to 20 knots faster than what would have been prudent for an airspeed in the traffic pattern at TEB. 5. The TRACON departure controller coordinated properly the flight of the DAS0 IFR traffie with the TEB control tower. 6. The TEB control tower coordinator failed to futlow preseribed proredures when he did not advise the local contrelier of the inbound DA50 traffic. 7 The clearance delivery controller failed io follow prescribed procedures when she did not advise the local controller of the DA5O. 8. There were several airplanes vperating at the airport and the radio ; frequency was congested. I 4 9 The local contro’ler was busy handling several airplanes and was I . surprised when the DA50 reported over the airport. 4 10, The local controller gave misleading traffic advisories concerning the direction of the transiting PA28, CF fm te wate Raa rn est Ant Ang kets ACESS FIN ELSE NTL ARSENE NT iS A SER CEP The DAO flighterew loss “situational awareness and probably misidentified the PA28 as the airplane they were to follow in the pattern for landing. The PA28 pilot probably did not see the DA50 until a moment before the collision. The radio frequency congestion interfered with the DA50 pilot-controller gommunication at a critical time in the sequence of events, The limitations to vision at night contributed to the DA50 captain's difficulty in resolving the collision course geometry in time to take appropriate action to avoid a collision. There was no clear poliey concerning VFR aircraft transiting the TER airport traffic area. 16. There was ineffective management and supervision over the TEB control tower controllers resulting in a failure to resolve operational problems. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of the accident was a breakdown Ir. air traffic control coordination which resulted in an air traffic conflict and the inability of the DA50 flighterew to "see and avoid" the other aireraft due to (1) an erroneous and inadequate traffic advisory and (2) the physiological limitations of human vision and reaction time at night. Air traffic control management contributed to the accident by fatling to insure that controllers were following prescribed procedures end by failing to recognize and correct operational deficiencies.

[ANALYSIS] This comment prompted the right turn made immediately by the first officer for separation. Also, the flightcrew would not have been surprised when they passed the PA28 and they would have assumed that it would no longer be a factor. Twenty-one seconds later, at 1720:39, the captain referred to another aircraft down low and the CVR recorded an unintelligible response from the first officer. The captain replied, "Beneath him," and the first officer stated, "I see him." It can only be presuined that the traffie sighted by both pilots in this Instance was the same airplane, N72BG, which was below the PA28. ‘Therefore, at 1720:41, when the controller asked the question "...you have the traffic, sir?" the captain acknowledged sighting the PA28 believed to be traveling westbound. However, the Safety Board does not believe the evidence supports this hypothesis because of the initial positions of the other three airplanes relative to the DASO and the indication that the flighterew still believed they were following a Twin Cessna at 1721. In addition, the controller admitted he did not know to which airplane the erew was referring - le stated it could have been the Twin Cessna, N1959T, or the PA28. If the DAS0 crew believed the PA28 was westbound, they would have expected the traffic to pass them quickly, which did not take place. Furthermore, the captain's comment, "You're eat 'em up," more likely implies the perception of an overtake situation rather chan convergence. Also, the belief that they were overtaking the PA28 would have been DSR RAPER STR OO BE SER ME BREET BE MORE TAPER OC, UTE NPS OE pe Lae RAN TRANS haat WP sr he VR tPees AP ly re deta: ~48- reinforeed if they heard the local controller tell the pilot of the PA28 that he was being overtaken by the DASO. In this hypothesis, the flighterew would have likeiy been confused sooner as to which airplane was the additional traffie and which airplane they were to follow on the downwind leg with the presence of N72BG, and they would not have taken an intervel on the PA28—at this point the DA50O was about 4.5 miles behind NJ.959T. Second Hypothesis.—This hypothesis assumes that the captain believed that the PA%8 was the alrcralt assigned as number two to land, that they were number three to follow, and that they had to space themselves with the PA28. However, it is assumed that the "westbound" report was not significant because the flighterew did not comment about head-on traffic at the same altitude, and the captain's comment, "You're eat ‘em up," implied the perception of an overtake situation. Therefore, the flightcrew discounted the two "westbound" reports made by the controller as a mistake.

[ANALYSIS] He sequenced the DA5O to be number three behind N1959T and correctly deseribed N1959'Ms location as turning downwind abeam the tower, However, the local controller did not provide the proper description for N1959T, a Twin Bonanza, to the flighterew. Furthermore, the controller erroneously reported the PA28 traffie as "westbound" in two instances 9 seconds apart ~ at 1720:05 and at 1720:14 and did not say that it was an overflight. (See figure 19.) At this time, N1959T was in view at about the 10 o'clock position slightly below the horizon about 1.2 miles away. The PA28 was in view just under a mile directly ahead. N72BG was about 30° to the right of the captain's viewing angle about 1.7 miles ahead and moving closer to a 1 o'clock position than the PA28. Actually, both N1959T and N72BG were turning to the downwind = ivecti- 4 at this time. Also, at 1720:14, the DA50 made about a 15° right bank. At 1720:26, when the local controller advised the pilot of the PA28 about the DA50 at his 6 o'clock position, all three airplanes were in view of the DA50 flighterew. (See figure 20.) At 1720:41, when the local controller asked the DA5O flighterew if they saw the traffic, N1959T was in view about 124° to their left almost 4 miles away traveling northbound on the downwind leg Uf the traffic pattern, the PA28 and N74BG appeared close together and in view about 40° to the left of their DER points. ‘The PA28 was about 3/4 mile ahead just above the horizon and N723G was slightly below the horizon about 1.3 miles away, and to the right of the PA28. (See figure 21.) About this time, when the captain acknowledged sighting what is believed he thought to be the "westbound" traffic, all three airplanes were in his view, and cnly N1959T was far to the left, out of the first officer's view.

[ANALYSIS] In addition to providing the eastbound overflight information, had the DA50 crew been given the identification of N1959T, the airplane they were supposed to follow, they would have been confronted with a different situation and it would have caused them to question the presence of the PA28. The Safety Board believes that this additional, but critical information, would have resulted in a different course of action by the crew which could have prevented the accident. 2.3 Piict_Actions The manner in which the DASO was flown was not in accordance with company training procedures regarding speed control and landing configuration sequence with respect to the approach being flown. The crew did not reduce thelr speed to 180 KIAS in a timely fashion as requested by the departure controller. Although the speed was decreasing, it was faster than it should have been under the circumstances regardless of what was permitted by regulation. The DA50 was about 163 KIAS over the airport and the speed was not down to about 140 KIAS until the DA50 passed the PA28. The increased speed resulted not only in a rapid closure rate with the PA28, but it also caused a late configuration of the DA50 for landing and caused it to preceed further away from the airport and outside the normal limits of the traffic pattern. The Safety Board noted that the captain cautioned the first officer to slow down on two occasions, at 1719:45 and at 1720:18, This indicates that the first officer may have been having difficulty managing airspeed and airplane configuration during the approach. ‘This difficulty would have distracted the captain from devoting more attention to the traffic situation. The Safety Board believes that although the flighterew were told they were following a Twin Cessna by the depurture controller, they should not have assumed this could be the case in the traffic pattern because VFR aircraft are mixed with IFR traffie for landing sequetice by the local controller at TEB. This apparent assurnption would have influenced their expectation of the traffic sequence in the pattern. Furthermore, the Safety Board believes that the excessive speed and the flighterew's preoccupation with the PA28 resulted in their losing awareness of the position of the DA50 with respect to the airport and the normal limits of the traffic pattern.

[ANALYSIS] Twenty-seven seconds later the ptlot of the PA28 reported to the local controller that he was 1 mile west of the airport (he was more than 2 miles), and he was instructed to report clear of the airport traffic to the "west." The local controller was busy working six airplanes and the radio frequency’ was congested. This mee it difficult for the DASO eaptain to contact the tower at CLIFO. By the time radio contact was established, the local controller had already cleared N72BG for a downwind departure and cleared N1959T to be number two to land when it reported overhead the airport. Since the N1959T's landing sequence was not provided by the controller until 1719:45, it is probable that the flighterew did not hear the controller's instruction to N1959T at this time because the captain was telling the first officer to stow it up at that mornent. Therefore, the Safety Board believes that the DASO flighterew may not have been aware that the local eoritrolier had sequenced N1959T between them and the Twin Cessna. At this point, the PA28 was almost directly ahead of the DA50 and 1.7 miles away. N1959T was 1.4 miles from the DA5O and about 43° to the left. The Safety Board further believes that the number of other airplanes the local controller was handling and the radio frequeney congestion that was generated detracted from his ability to prevent a conflict between the PA28 and the DASO when it made its initial call, Based on the radar data and cockpit visibllilty study at 1719:45, when the DAS0 was about a mile west of the airport, the Twin Bonanza, N19$9T, and the PA28 were passing over the airport in front of and in the DASO flighterew's view. (See figure 17.) N1950T was about 40° to the left at 900 feet and the PA28 was about 20° to the left of the crew's DER at 1,600 feet on the horizon, The Beech Baron, N72BG, which had derarted to the east after takeo?? from runway 19, was located slightly to the left of the crew's DER, and to the right of the PA28 and slightly below the horizon at 350 feet.

[ANALYSIS] Twenty seconds later, both the PA8 and N72BG maintained their approximate positions relative to one another and moved to the left horizontally about 46°. N1959T moved horizontally to the left and out of the first officer's view. (See figure 22.) At 1721:00, when the first officer asked “What kind of Cessna is that?" the PA28 and N72BG remained in view in about their same relative positions about 80° to the left of the crew's DER, but the PA28 was about 1/2 mile away and N72BQ was about 1.6 miles away. Eight seconds before the collision when the captain said, "Hey, wateh out, this guy's comin’ right at us," the DA50 was in about a 30° left bari and the PA28 and N723G were about 63° to the left and about 20° above the DER i + the captain, They were not in the first officer's view. The PA28 was just above the horizon ancl N72BQ was directly below the PA28 which was about 1/3 mile away or about 2,906 feet from the DA50. N1959T was out of the view of both pilots. (See figure 23.) About 30 seconds before the collision, the DA50 was positioned about 109° to the right and about 1° below the DER for the PA28 pilot. Eight seconds before the collision, the DA50 was about 77° to the right and 1/2° below the DER, The DASO captain's acknowledgment of having observed th: ivaffie at 1720243 and the controller's instruction for the flight to "maintain visuat" enced any further alr traffic control efforts to sequence the traffic. This was considered a proper action based on preseribed ATC procedures. However, the local cortroller's failure to mention that the additional traffic was an eastbound overflight instead of "westbound" is considered to be a factor in the accident.

[ANALYSIS] Under these conditions it takes longer for a pilot to detect conflicting traffic, make a decision, and take evasive action. The FAA's estimate of 12.5 seconds for such action to take place does not take Into gecount the extended time required at night. When the captain said, "Hey, watch out. This guy's coming right at us," it probably took longer than 6 seconds to detect the PA28 and perceive a collision course. Four seconds later, he decided on an evasive maneuver and told the first officer to "go down." However, as the airplanes came closer, 2 seconds Jater he must have realized that his initial decision was not the best escape maneuver and told the first officer, "Naw, go up." At this last command, there was probably 1.6 seconds remaining before the ¢ollision. With the first officer flying, overall response time wouid have been further delayed because of the additional time required for him to react. Also, given the airplane's response time, therc was no time to escape. The angle of bank to the right, as indicated by the physical evidence was likely the result of the pilots' instinctive reaction to turn away from the PA28. Why the captain did not take control of the airplane when he perceived the collision threat is not known. 2.4 Air Traffic Management Because of the location of the TEB airport with respect to LaGunrdia and Newark International Airports, the design of the TCA, and the air traffic congestion in the area, pilots flying general aviation airplanes for pleasure under VFR are inclined to overfly TEB in order to take the shortest route to the Hudson River for a seenie flight. In fact, under the present design the Hudson River VFR corridor is frequently used by these pilots as a major transition to remain clear of controlled airspace and there are no rules specific to operations through the corridor. No permission is needed to fly in the corridor and radio contact with a controlling agency ts not required. Furthermore, because of the heavy IFR traffic demands on ATC in the New York area, pilots flying under VFR are no doubt unsuccessful in obtaining radar traffic advisory services in the area at every request. Such a situation creates a critical need for clear a critical need for clear and concise communications, ATC coordination, and the necessity for pilots to exercise vigilant "see and avoid" practices when landing and departing the TEB airport. The importance of these factors are demonstrated in this accident. A breakdown in communications occurred within the TEB control tower. This breakdown underscores the fact that, in spite of the redundancies designed into the system of ATC, deviations from standard operating procedures and practices can result in accidents. Several factors led to the breakdown {{n communications. In view of the testimony from FAA personnel, it was evident that they did not place much emphasis on the importance of the flight progress strip in this particular accident involving a VFR tower.

[ANALYSIS] Although the FAA currently plans to require TCAS only on 14 CFR Part 121 alreraft, {{t is believed that some operators of well-equipped corporate and general aviation aircraft will also purchase TCAS. An operating TCAS would undoubtedly have displayed a caution and warning based on information from the Mode C transponder on N1977H, and also given the crew of N784B a recommended avoidance maneuver. 2. ANALYSIS 2.1 General Both airplanes were maintained in accordance with prescribed Federal . I regulations. There was no evidence of any discrepancies in either airplane that would - have had a bearing on the accident. . 4 The DA50 flighterew were certificated and currently qualified to fly the a DA50; both were experienced pilots. The fact that the first officer was flying the a airplane was consistant with the company's policies and procedures. Postmortem {{ examinations and other evidence disclosed no medical factors which would have affected e adverse’ their ability to operate the airplane. The pilot of the PA28 was certificated and currently qualified to fly the airplane. He was a less experienced pilot with 289 hours of flight time. Postmortem examination and other evidence revealed no medical factors which would have detracted from his physical ability to operate the airplane. 67 For more detalled information read, Aircraft Accident Report--"Wings West Airlines Beech C99, N6309U, and Aesthetec, Inc., Rockwell Commander 112TC, N112S™ near San Las Obispo, California August 24, 1984" (NTSB/AAR-85/07), The New York TRACON controller and the four controllers in the TEB control tower involved with the handling of the DA50 were full performance level controllers and qualified according to existing Federal regulations and FAA policies and procedures. STROMA IT AAD NATO I RON LE LSP SERIA RE CLS it san Y In review of the faets, conditions, and the circumstances in the accident, the Safety Board considered the following elements in its analysis of tne aceident: (1) the air traifie coordination and management which resulted in the local controller being unaware of the DA5O's inbound flight; (2) the complex air traffie situation created by the design of . & the TCA with respect to the TEB airspace and the authorization of VFR overflights at . e conflicting altitudes, and the mix between low and high performance general aviation aiveraft; (3) the contribution of unclear ov misleading traffic advisories; (4) the reliance on achieving visual separation with the use of the "see and avoid" concept in a high density air traffic airspace; and (5) the performance of the pilots In the conduct of their flights. ride et aE Ey NL PP ELS Se row bi ARN AN ate EE me atte Bane Moat sy 2.2 ' The Accident The accident sequence of events began when the ARTS computer did not automatically acquire the DA50 after it departed Morristown because it left about 1 minutes earlier than proposed on the flightplen.

[ANALYSIS] The Safety Board believes that the local controller should have been more aggressive in controlling the traffic instead of attempting to accommodate ail of the traffic on request. Furthermore, the Safety Board believes that had the control tower been equipped with a BRITE IV radar display enhanced with alpha-numeric capability, the local controller could have detected the arrival of the DA50 in advance and taken action to provide timely advisories without the use of a flight strip. Such enhancement of the BRITE display at TEB would further improve the situation found as the result of a previous midair collision near TEB. 7/ In addition, the TEB control tower is limited to providing VFR service only because the controllers are not radar qualified and 17 For more detalled informaton read, Aircraft Accident Report--"Ronson Aviation, Bell 206B, N27670 and Seminola Air Charter, Piper PA-34-200T, N8110R, Midair Collision, East Rutherford, New Jersey, September 23, 1981" (NTSB-AAR-82-6). Serer tees ST “RTE sin ATENEO NURS TENET” AA OTT if a andy ak dich ds ees pt Be sien EPA cat Sy Pelee tanec tlied WIR lage Pa AGB Siar Labia gd aS Beg Siac aia Met Laeaecrt stort ry =5$- cannot provide positive separation and the tower is not designed to provide this service. Reliance on the "see and avoid" concept as a sole means of providing air traffic separation is questionable, in view of the compressed airspace situation and the mix of aircraft types : E with significant performance differences in airspeeds and maneuvering capabilities. ‘ PI The Safety Board concludes that the ATC system designed to provide traffic F x advisories to aireraft operating within the TEB terminal area broke down. Because of the j _ airspace structure and the cumulative deficiencies that were occurring at TEB, the Safety E 4q Board believes that the tower manager as well as regional ATC management should have ‘ a recognized the need for corrective action before this accident occurred. Failure on their a part to identify and correct these problems suggests a system failure at TEB which i s created the environment for the multiple individual failures that led to the accident and that reflect poorty on the overall managerial policy and practices at the control tower. : Consequently, the Safety Board believes that the lack of aggressive preventive measures by the FAA ATC directly contributed to the cause of the accident. 3. CONCLUSIONS : 3.1 Findings NN ee ES Pe Fe ee OD 1, The pilots {{n both airplanes were qualified for their respective flights and were familiar with the operations at TEB. There were no apparent medical factors affecting their performance. 2, The controllers were qualified as fult performance level controllers.

[ANALYSIS] If one was not, it was his responsibility to prepare one before control of the DA30 was transferred to the TEB local controller. The coordinator's dectsion to ask the Clearanve delivery controller to cover for him while he was absent from the cab was proper and routine. However, he # }} did not give a relief briefing to the clearance delivery controller, nor did she request a — briefing. The clearance delivery controller should have been made aware of all the active traffic handled by the coordinator. ‘The lack of 4 proper briefing on the part of both —. controllers was contrary to required procedures and precluded a second opportunity to the 4 controllers to stop the sequence of events that led to the acvident. 4 20 aR SE EEN CTS ORD WSIS SEEN ORS FBO EY AOE FIN EE PUONESER BHA RAS — ’ oH reese es aay Sopa detimigsie sy see ns ay SS a mS are ~33~ The clearance delivery controller also failed to insure that the local controller was aware of the DASO. If the ground sontrolier had moved a strip over to the local controller, it would be reasonable to expect the clearance delivery controller to think that the ground controller heard her report and provided a strip to the local controller at that ‘nstant. Without physically positioning the strip herse:f or without receiving verbal vonfiguration from either controller, she would have had no way of knowing that the strip pertained to the DASO. In fact, the strip the ground controlier mada up was not prepared until about 2 minutes later when the airplane was over the airport — too late to have permitted the clearance delivery controller to provide adequate sequencing by issuing advisories. In addition, the Safety Board believes that a breakdown in coordination occurred in the control tower which set the stage for the accident. In the Safety Board's opinion, the local controller may have been alerted sooner to a potential conflict between the DASO and the PA28 if the coordination process had been timely, accurate, and complete. At 1714, the DA5O flighterew knew that they were following a Twin Cessna to TEB. About 3 minutes later, they were about 5 1/2 miles from the airpert and were told that the traffic they were following was over the airport. This traffie was the Twin Cessna (N68734) and it was not considered a factor in the ancident because of its distance from the other airplanes. At that time, the PA28 was about 30° to the right of and about 31/2 miles away from the DA50 and abcut the same distance away from the airport. Twenty-seven seconds later the ptlot of the PA28 reported to the local controller that he was 1 mile west of the airport (he was more than 2 miles), and he was instructed to report clear of the airport traffic to the "west." The local controller was busy working six airplanes and the radio frequency’ was congested. This mee it difficult for the DASO eaptain to contact the tower at CLIFO.

[ANALYSIS] Also, her attention was divided between crosschecking the takeoff performance numbers and visually clearing the runway, reducing the amount of time she devoted to inspecting the external environment. It is likely that, due to her divided attention, she performed a cursory visual scan of the runway environment and made the associated routine callout. Thus, workload, multitasking, time pressure, and expectancy likely played a role in the FO not detecting the captain’s surface navigation errors between the time the airplane passed taxiway K and the time the airplane taxied onto runway 4L. The FB recalled that his visual attention was focused inside the flight deck from the time the crew received the load closeout until the airplane crossed the holding position markings and entered runway 4L. During that period, he recalled handing the printed load closeout to the other crewmembers, cleaning up paperwork on the center pedestal, and adjusting his radios. He recalled hearing some confusion about crossing a runway and looking up, at which time the airplane was across the runway 4L hold short markings. As a result, he would no longer have been able to see the signs for runway 4L. Thus, the FB did not detect the captain’s deviation from AAL106’s taxi clearance because he was performing operational tasks other than looking outside and monitoring the airplane’s position. Aviation Investigation Report AIR-24-01 39 The NTSB concludes that the FO and FB were likely distracted from their primary duty of assisting the captain in safely taxiing the airplane by other operational activities, which resulted in the crew’s loss of situational awareness during a critical phase of flight. 2.2.1.3 American Airlines Crew’s Nondetection of DAL1943 on Runway 4L DAL1943 was positioned for takeoff on runway 4L for about 2 minutes before beginning its takeoff roll, and began its takeoff roll about the time AAL106 started turning onto taxiway J. As discussed in the previous section, the AAL106 captain and FO recalled visually scanning the runway, but neither recalled seeing DAL1943 on the runway. The captain and FO recalled scanning the runway at some point during the 11 seconds between the time the captain began turning the airplane onto taxiway J and the time the airplane crossed the hold short markings and the RELs activated. The crew recalled that the captain also switched on some additional exterior lights during this time, which briefly required some of his visual attention inside the cockpit. As the airplane began to turn onto and align with taxiway J, the approach end of runway 4L was to the captain’s right and may have been partially obstructed by the FO’s body and aircraft structure. Thus, it is possible the captain would not have had an opportunity to see DAL1943 due to his obstructed view. Examination of satellite images, illumination diagrams, and on-scene nighttime photographs of the airport suggest that a significant amount of airport lighting was present between the two airplanes in various colors and intensities.

[CONCLUSIONS > FINDINGS] Kennedy International Airport was operational at the time of the incident and functioned as designed, generating both aural and visual alerts when American Airlines flight 106 crossed runway 4L while Delta Air Lines flight 1943 was departing, and likely reduced the severity of the incident by preventing a runway collision. 10. The ground controller expected the American Airlines flight 106 crew to adhere to the assigned taxi instructions and did not detect the flight crew’s surface navigation error and subsequent turn onto taxiway J because he was performing a lesser priority task that entailed looking down. 11. The local controller acted in a timely and appropriate manner following the airport surface detection equipment model-X alert by cancelling Delta Air Lines flight 1943’s takeoff clearance. 12. The John F. Kennedy International Airport air traffic control tower team had the responsibility of scanning the runways and airport environment but did not effectively prioritize their duties to ensure a continuous scan, which resulted in their nondetection of the American Airlines flight 106 crew’s deviation from taxi instructions. 13. Cockpit voice recorders (CVRs) with a 25-hour recording capability are necessary because valuable information continues to be overwritten on CVRs that are designed to record only 2 hours of audio data. 3.2 Probable Cause The NTSB determines that the probable cause of this incident was the American Airlines flight 106 (AAL106) crew’s surface navigation error due to distractions caused by their performance of concurrent operational tasks during taxi, which resulted in a loss of situational awareness. Contributing to the incident was the air traffic control tower team’s nondetection of the AAL106 crew’s deviation from taxi instructions while performing concurrent operational tasks; the timing of the runway status light system, which activated too late to prevent the AAL106 crew from crossing the runway hold short line; and American Airlines’ lack of adequate risk controls to prevent concurrent flight crew tasks from leading to distraction, loss of situational awareness, and deviation from an authorized taxi clearance. Reducing the severity of the incident, and likely preventing an accident, was the activation of the ASDE-X warning in the air traffic control tower and the local controller’s prompt cancellation of DAL1943’s takeoff clearance. Aviation Investigation Report AIR-24-01 54

[ANALYSIS] Believing that AAL106 was conducting the taxi as instructed, he did not ensure that the flight crew completed the right turn onto taxiway K before he began attending to other duties. FAA Order 7110.65Z, Air Traffic Control, paragraph 2−1−2, Duty Priority, stated that controllers were to give first priority to separating aircraft and issuing safety alerts as required. Therefore, the NTSB concludes that the ground controller expected the AAL106 flight crew to adhere to the assigned taxi instructions and did not detect the flight crew’s surface navigation error and subsequent turn onto taxiway J because he was performing a lesser priority task that entailed looking down. 2.5.2 Local Controller The local controller stated that he performed a visual scan before issuing the takeoff clearance to DAL1943, then he started talking to other airplanes that were taxiing to line up and wait, one on runway 4L and another on runway 31L. Aviation Investigation Report AIR-24-01 50 Shortly after the controller provided these instructions, the ASDE-X alerted. He looked at the ASDE-X display and saw the aircraft involved were DAL1943 and AAL106. The local controller cancelled DAL1943’s takeoff clearance 5 seconds after the ASDE-X alerted. He repeated the instruction a second time, and the DAL1943 crew responded that they were rejecting the takeoff. The NTSB concludes that the local controller acted in a timely and appropriate manner following the ASDE-X alert by cancelling DAL1943’s takeoff clearance. 2.5.3 Air Traffic Control Tower Team At the time of the incident, the tower was changing the runway configuration, which required some of the controllers to physically move into different positions in the tower, as well as change all their equipment to the new configuration. Airport operations was on the airfield conducting checks and switching equipment for the runway change as well. On the night of the incident, the runway configuration change was requested to be completed earlier than the time it typically took place between 2130 and 2230. Although a runway configuration change is an important priority, and, on the night of the incident, occurred at an atypical time, it is a routine event that does not always occur at prescribed times. The whole tower team has the primary responsibility of ensuring safe and effective operations in the tower and to perform continual visual scans of the runways. In this case, the cab coordinator, flight data/clearance delivery controller, and operations supervisor were busy making preparations for the runway change and were not giving their full attention to the air traffic operations at the time. The tower team, collectively, did not effectively prioritize their duties by ensuring that the team maintained a continuous scan of the airport environment as preparations for the runway change were performed.

[ANALYSIS] The unexpected tasks included dealing with multiple weather advisories from dispatch via ACARS and informing crewmembers accordingly. The tasks with timing determined by external events included those tasks triggered by the receipt of weather advisories and tasks triggered by receipt of the load closeout, such as crosschecking final weights and V-speeds. Aviation Investigation Report AIR-24-01 37 An alternative framework from the scientific literature on human error involves branching error, which can occur when a task involves an action schema that is associated with more than one possible end goal. In such situations, periodic attentional checks are needed to detect one’s arrival at a choice point. This is consistent with the previous discussion about the captain’s prospective memory error. In a branching error, this check does not occur, behavior becomes disconnected from intention, and actions beyond the choice point become driven by the schemas that are commonly performed in the new situational context. This latter form of a skill-based action slip is known as environmental capture (Reason 1990). According to this framework, lower-level schemas caused the captain to continue the taxi along taxiway B after he missed the turn at taxiway K. Shortly thereafter, the airplane was directly facing taxiway J and the highly salient flashing yellow guard lights for runway 4L. A typical behavior pattern performed in that location involved taxiing onto taxiway J and crossing runway 4L (the standard route to runway 31L). In the absence of focused attention, schema theory suggests that the captain’s actions may have become disconnected from his intentions and situational cues could have prompted him to continue along a common, but incorrect, route. This would explain the captain’s subsequent surprise and confusion when the ground controller alerted him to the discrepancy between AAL106’s clearance and the airplane’s position. Thus, a branching error appears to be a good fit for the captain’s continuation of the taxi onto taxiway J and across runway 4L. The NTSB concludes that the captain’s deviation from AAL106’s taxi clearance likely resulted from several factors, including an early clearance to cross runway 31L, interruptions and multitasking related to the crew’s delayed receipt of the load closeout, the captain’s prospective memory error in forgetting to turn right at taxiway K, and environmental capture, which prompted the captain to proceed along a familiar, but incorrect, route. 2.2.1.2 American Airlines FO and FB’s Nondetection of the Captain’s Error The NTSB sought to understand why neither the FO nor the FB detected the captain’s error between the time the airplane passed taxiway K until the RELs activated. The FO told investigators she was head-down dealing with tasks related to the load closeout for the approximate 1.5 minutes from the time the load closeout was received until the captain began turning onto taxiway J and toward runway 4L. The missed turn at taxiway K occurred during this time that the FO’s visual attention was likely focused inside the flight deck.

[ANALYSIS] Thus, it is possible the captain would not have had an opportunity to see DAL1943 due to his obstructed view. Examination of satellite images, illumination diagrams, and on-scene nighttime photographs of the airport suggest that a significant amount of airport lighting was present between the two airplanes in various colors and intensities. Due to flattening of perspective that occurs when viewing distant objects on flat terrain, DAL1943, which was around 3,000 ft away from AAL106’s position, likely appeared as a small target within a dense array of surrounding lights that varied in color and brightness, including flashing yellow lights at each taxiway/runway intersection; blue taxiway edge lights; green taxiway centerline lights; green and yellow taxiway lead-on lights; yellow lighted taxiway signs; and red, white, and yellow edge lights for runway 4L. These surrounding lights would have been distractors during a visual search, making a target more difficult to identify. Additionally, DAL1943 was moving directly toward AAL106, and would have had little apparent motion, making it less likely to attract focused visual inspection. Given the fact that the FO did not see DAL1943 before entering the runway, it is likely that she never fixated directly on the airplane. Thus, the FO’s nondetection of DAL1943 likely resulted from limitations of the human visual system for detecting small stationary targets in a complex visual scene in a limited time. Even if the crew had noticed the presence of DAL1943, it might have been difficult for them to determine whether the airplane was on runway 4L until after Aviation Investigation Report AIR-24-01 40 AAL106 had entered runway 4L and its crew could look straight down the runway. In fact, this is when the FB first recalled seeing DAL1943. It is likely that the forwardfacing white taxi/landing lights were the most salient external lights on DAL1943 when viewed from the perspective of AAL106. Another potential explanatory factor is expectation bias, a cognitive bias that reduces the likelihood a person will notice a mismatch between what they expect to see and the actual state of things. In this event, the crew believed that the controller had cleared them to cross the runway in front of them. The AAL106 captain and FO had likely performed thousands of visual scans of runways in the past, and it is likely that most of these scans revealed no conflicting aircraft. Therefore, the crew probably did not expect another airplane to be on the runway and moving toward them, which made them vulnerable to inspecting the runway and making related callouts in rote fashion rather than systematically searching for a small target. Thus, expectation bias could explain how the captain and FO did not observe DAL1943 before entering runway 4L.

[ANALYSIS] While his flight was in Waimea Canyon and encountered deteriorating weather at the north end, he would have been unable to see the weather conditions north of the canyon until his tour exited the canyon over the west rim. Aviation Investigation Report NTSB/AIR-22-05 50 Upon exiting the canyon, the accident pilot may have optimistically thought that any encounter with poor visibility along the high elevation canyon rim would be brief and that conditions would improve as the flight continued toward the Nāpali Coast, where he could descend the helicopter over the terrain downslope, increasing the flight’s clearance below the clouds. However, due to the atypical weather pattern on the day of the accident, such a supposition would have been incorrect because some of the clouds moving in from the northwest were very low, extending to the ground as fog in some areas as they moved over the terrain on the northwest side of the island. Thus, the accident pilot’s decision to continue the flight into an area of deteriorating weather at the north end of Waimea Canyon and then continue farther north of the canyon may have resulted from a lack of relevant preflight weather information, which hindered his ability to develop an accurate mental model of the atypical weather situation. Research on pilots’ weather assessments has shown that some pilots may also continue visual flight into adverse weather conditions due to deficiencies in situation assessment. That is, they do not adequately recognize when the conditions that they can see in front of them have deteriorated to the point that they are unsafe. One simulator study found that only 10 out of 32 VFR pilots made a timely decision to divert their flights when they encountered IMC. Further, pilots in the study varied substantially in their estimates of ceiling height and visibility (and tended to overestimate both), and those who continued the flights made less accurate (higher) estimate of visibility than those who diverted (Goh and Wiegmann 2001). Another study that presented pilots with videos of in-flight weather conditions found that their assessments of the conditions were inaccurate and overly optimistic (Coyne et al. 2008). Thus, it is possible that the accident pilot’s decision to continue the VFR flight into deteriorating weather was influenced, in part, by an inaccurate in-flight assessment of the weather conditions. Another possibility is that the pilot accurately assessed the conditions but still chose to fly into an area of deteriorating weather, which would indicate a deficiency in the action selection stage of decision-making.

[ANALYSIS] A pilot’s ability to manually control a helicopter during VFR flight operations is highly dependent on the availability of good outside visual references. Maintaining such visual references is particularly critical for pilots of helicopters not certified for instrument flight rules (IFR) operations (like the accident helicopter and most helicopters used for air tours), which are not typically equipped with features designed to help a pilot to maintain orientation and helicopter control in IMC.77 At a minimum, these include instruments, trim, and stability enhancements, which may include a stability augmentation system (SAS) or autopilot, or both.78 Historically, the primary strategy for preventing helicopter accidents resulting from inadvertent encounters with IMC has relied on pilots to avoid entering reduced visibility conditions. As discussed in previous sections of this report, pilot training interventions and improved safety assurance can help reduce inadvertent IMC encounters by improving pilot weather assessment and decision-making skills as well as monitoring the effectiveness of existing procedural and regulatory mitigations. However, the continuing occurrence of VFR-into-IMC accidents demonstrates that this strategy alone is not sufficient. On the day of the accident, the accident pilot and at least three other tour pilots flew their helicopters into reduced visibility weather conditions. This suggests that such scenarios are not rare, at least for air tour operations in Hawaii where rapidly changing weather conditions may be more difficult for a pilot to anticipate. Another strategy to increase the likelihood that helicopter pilots can successfully avoid or escape inadvertent IMC encounters involves the increased use of helicopter safety technology. USHST noted that helicopter trim and stability are increasingly important as visual conditions deteriorate, asserting that “if all rotorcraft are designed to meet some of the IFR stability requirements, many [in-flight-loss-ofcontrol] accidents could be avoided, as the aircraft stability would help the pilot 76 The project reviewed 31 accidents with sufficient flight data to perform the analysis. Although five of the accident helicopters were equipped for instrument flight rules (IFR) operations, all were operating under VFR, and three of the accidents occurred at night (increasing the difficulty for the pilot to reacquire visual references after inadvertently entering IMC). 77 IFR refers generally to operations in weather conditions that are below the minimum required for VFR flight. 78 Equipment requirements for normal category rotorcraft certified for IFR flight are specified in 14 CFR Part 27, Appendix B. Aviation Investigation Report NTSB/AIR-22-05 87 maintain positive control during temporary losses of visual cueing or disorientation” (USHST 2021a). The NTSB believes that such technology (when combined with the use of instruments and a terrain awareness display) may also help reduce CFIT accidents by reducing the pilot workload associated with maintaining helicopter control in reduced visibility, allowing the pilot to allocate more attention to terrain avoidance while maneuvering to escape the conditions.

[ANALYSIS] Also, the Company 2 pilot entered reduced visibility conditions north of Waimea Canyon before taking action to divert, and he subsequently entered areas of visible moisture along the Nāpali Coast after approaching it from the east. Onboard video from the Company 1 and Company 2 helicopters showed that, when these helicopters reached the north shore, areas of clearer weather (for example, east of Hanalei Bay) were available for them to continue east along a clockwise tour around the island, but they instead flew west. If such continuation of flights into reduced visibility weather conditions were indicative of routine noncompliance in everyday decision-making, then such routine noncompliance could have led to a redefinition of “acceptable” weather conditions among these pilots that was inconsistent with FAA requirements. This could explain how an experienced pilot who was not widely regarded as a risk-taker could find it Aviation Investigation Report NTSB/AIR-22-05 53 acceptable to continue a flight into below-minimum weather conditions, however brief he may have expected this encounter to be. In summary, factors relevant to the accident pilot’s decision to continue the VFR flight into deteriorating weather conditions include a lack of relevant weather information, an atypical weather pattern, and the possibilities that he inadequately assessed the weather in-flight, was overconfident in his abilities, or could have been influenced by a drift toward risky operating practices among air tour pilots operating on the island of Kauai. However, due to conflicting or insufficient evidence, it is difficult to assess the role or the relative importance of the various possible weather assessment and risk perception bias considerations.41 Thus, the NTSB concludes that the pilot’s decision to continue the flight into deteriorating visibility was likely influenced by a lack of relevant weather information and an atypical weather pattern and may have been influenced by the possibilities that he inadequately assessed the weather conditions in flight or was overconfident in his abilities. The NTSB further concludes that, considering that at least three other tour flights entered reduced visibility conditions on the day of the accident, it is possible that procedural drift toward risky weather-related operating practices existed among pilots of the local air tour community. 2.3.2 Lack of Effective Cue-Based Weather Training The NTSB has long advocated for the FAA to develop and require cue-based weather training for Hawaii air tour pilots to assist them in their ability to accurately assess hazardous local weather conditions and make appropriate in-flight weather-related decisions. Such training, if offered on a recurrent basis, may help prevent procedural drift and normalization of deviance among pilots by periodically recalibrating their perceptions of what constitutes acceptable weather. We first expressed our interest in cue-based weather training in 2007 in our report on another helicopter air tour accident involving VFR flight into IMC on Kauai.

[CONCLUSIONS > FINDINGS] If all commercial air tour aircraft were required to be equipped with crash-resistant recording systems that capture images, flight deck audio, and flight data, improved information about accident circumstances would permit more definitive evaluation of the causes of fatal air tour accidents and identification of more effective measures to prevent these accidents. 18. Although Airbus Helicopters had developed and issued service bulletins for the voluntary retrofit of recording systems on Airbus AS350 helicopters more than 2 years before the accident, Safari Aviation Inc. had not installed such recorders in the accident helicopter or any other helicopter in its fleet. 19. Increased voluntary adoption of helicopter safety technologies designed to help reduce accidents resulting from inadvertent encounters with instrument meteorological conditions can help save lives. 20. The use of simulation technologies in helicopter pilot training has the potential to help reduce accidents involving inadvertent encounters with instrument meteorological conditions by enhancing the weather-related decision-making and assessment aspect of scenario-based training and Aviation Investigation Report NTSB/AIR-22-05 96 providing opportunities for pilots to develop skills to detect and recover from vestibular illusions that could lead to spatial disorientation. 3.2 Probable Cause The NTSB determines that the probable cause of this accident was the pilot’s decision to continue flight under visual flight rules (VFR) into instrument meteorological conditions (IMC), which resulted in the collision into terrain. Contributing to the accident was Safari Aviation Inc.’s lack of safety management processes to identify hazards and mitigate the risks associated with factors that influence pilots to continue VFR flight into IMC. Also contributing to the accident was the Federal Aviation Administration’s delayed implementation of a Hawaii aviation weather camera program, its lack of leadership in the development of a cue-based weather training program for Hawaii air tour pilots, and its ineffective monitoring and oversight of Hawaii air tour operators’ weather-related operating practices. Aviation Investigation Report NTSB/AIR-22-05 97 4. Safety Recommendations 4.1 New Recommendations As a result of this investigation, the National Transportation Safety Board makes the following new safety recommendations.

[CONCLUSIONS > FINDINGS] Therefore, due to the relevance of these safety recommendations to the circumstances of this accident, the NTSB reiterates Safety Recommendations A-21-5 and -6. Aviation Investigation Report NTSB/AIR-22-05 93 3. Conclusions 3.1 Findings 1. None of the following safety issues were identified for the accident flight: (1) pilot qualification deficiencies or impairment due to medical condition, alcohol, other drugs, or fatigue; (2) helicopter malfunction or failure; or (3) pressure on the pilot from other Safari Aviation Inc. managers to complete the flight. 2. The accident pilot continued the tour flight into an area of deteriorating weather until he encountered instrument meteorological conditions and lost adequate visual references. 3. Due to the lack of information about the helicopter’s flight track or any other parameters, it could not be determined whether the accident pilot maintained controlled flight after encountering instrument meteorological conditions or lost control of the helicopter before it struck terrain. 4. The pilot’s decision to continue the flight into deteriorating visibility was likely influenced by a lack of relevant weather information and an atypical weather pattern and may have been influenced by the possibilities that he inadequately assessed the weather conditions in flight or was overconfident in his abilities. 5. Considering that at least three other tour flights entered reduced visibility conditions on the day of the accident, it is possible that procedural drift toward risky weather-related operating practices existed among pilots of the local air tour community. 6. As a result of the Federal Aviation Administration‘s (FAA) lack of leadership and expert guidance in developing cue-based weather training for air tour operators in Hawaii, Safari Aviation Inc.’s pilot training program, which was approved by the Honolulu flight standards district office, did not provide Safari’s pilots with the type of training that the FAA’s cue-based training research determined was effective for improving pilots’ skills for accurately assessing and avoiding hazardous in-flight weather conditions. 7. Although Safari Aviation Inc. had a policy that defined company expectations for pilot adherence to minimum weather requirements, it did Aviation Investigation Report NTSB/AIR-22-05 94 not have adequate safety assurance processes to assess whether company strategies to reduce pilots’ risk of inadvertent encounters with instrument meteorological conditions were effective. 8. The Federal Aviation Administration’s full implementation of a Hawaii weather camera system as planned will provide access to continuously updated weather imagery that will support pilots’ weather-avoidance decision-making and likely reduce weather-related air tour accidents in the state. 9. Now that the first aviation weather cameras are operational in Hawaii, the ability for en route pilots to obtain essential information from ground support personnel, such as flight service station specialists, based on the cameras’ imagery is particularly relevant due to the rapidly changing weather conditions on the islands. 10.

[ANALYSIS] Although Airbus Helicopters had developed and issued service bulletins for the voluntary retrofit of recording systems on Airbus AS350 helicopters more than 2 years before the accident, Safari had not installed such recorders in the accident helicopter or any other helicopter in its fleet. Therefore, the NTSB reiterates Safety Recommendation A-13-13. 2.8 Emerging Technologies for Preventing Accidents Involving Inadvertent Encounters with IMC 2.8.1 Helicopter Safety Technologies Reviews of NTSB accident data by the US Helicopter Safety Team (USHST) have repeatedly identified inadvertent flight into IMC as one of the leading categories of fatal helicopter accidents (USHST 2021b).75 An inadvertent encounter with IMC in a helicopter is considered an emergency that often leads to a fatal outcome, such as an in-flight loss of control or CFIT accident (FAA 2019a). A USHST analysis of fatal helicopter accidents involving such encounters found that the average time between 75 The USHST is a volunteer group of US government and industry stakeholders formed in 2013 to improve the safety of civil helicopter operations. Its efforts include analyzing NTSB helicopter accident data, assigning a single occurrence category to best characterize each event, and using the results of its analysis to prioritize intervention strategies to reduce fatal accidents (USHST 2017). The USHST’s most recent analysis included data from 198 fatal helicopter accidents between 2009 and 2018. Although this review ranked inadvertent flight into IMC (also referred to internationally as unintended flight into IMC) as the third-most common category, the USHST noted that inadvertent flight into IMC may be the precursor to accidents involving in-flight loss of control (which ranked first), low-altitude operations (which ranked second), or CFIT (which ranked fifth) (USHST 2021b). During a previous review (which included data between 2009 and 2013), inadvertent flight into IMC ranked second (USHST 2017). Aviation Investigation Report NTSB/AIR-22-05 86 IMC entry and ground impact was just 56 seconds (USHST 2021b).76 This is due, in part, to the inherent instability of helicopters. A pilot’s ability to manually control a helicopter during VFR flight operations is highly dependent on the availability of good outside visual references.

[PROBABLE CAUSE] Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the pilot’s decision to continue flight under visual flight rules into instrument meteorological conditions, which resulted in the pilot’s spatial disorientation and loss of control. Contributing to the accident was the pilot’s likely self-induced pressure and the pilot’s plan continuation bias, which adversely affected his decision-making, and Island Express Helicopters Inc.’s inadequate review and oversight of its safety management processes. NTSB Aircraft Accident Report vii Safety Issues The investigation evaluated the following safety issues: • The pilot’s preflight weather and flight risk planning. The pilot completed a flight risk analysis form about 2 hours before the accident flight’s departure. Based on the form’s risk scoring criteria, the pilot’s score for the accident flight was within the company’s low-risk category. Updated weather information available at the time the accident flight departed included conditions that met the criteria for the form’s risk items that would have required the pilot to seek input from the director of operations and to provide an alternative plan. However, company guidance was unclear as to whether the accident pilot was expected to complete an updated form (and he did not do so). • The flight’s entry into instrument meteorological conditions (IMC) and the pilot’s inadequate adverse weather avoidance. At the time the pilot took action to initiate a climb, the helicopter had already begun penetrating clouds. Although the pilot’s adverse-weather-avoidance training emphasized avoiding entry into IMC by slowing the helicopter and maneuvering or landing, there was no evidence that he attempted to do so. • The pilot’s spatial disorientation. As the helicopter climbed rapidly into the cloud layer and IMC while in a gradual left turn, the pilot’s associated loss of outside visual references made him susceptible to experiencing vestibular illusions (in which the vestibular system in the inner ear produces a false sense of helicopter attitude and trajectory) that can lead to spatial disorientation. • Influences on the pilot’s decision to continue the flight into adverse weather. The pilot’s continuation of the accident flight into IMC was inconsistent with his typical judgment and decision-making behavior and was likely influenced by his self-induced pressure, lack of an alternate plan, and plan continuation bias. • Island Express’ incomplete implementation of its safety management system (SMS). The company had an SMS that was neither required by the Federal Aviation Administration (FAA) nor part of the company’s FAA-approved or -accepted programs. Although the company used some SMS tools, it did not implement the entire program and did not perform any safety assurance evaluations, such as those that could have ensured the effectiveness of the flight risk analysis forms. • The benefits of a mandatory SMS.

[ANALYSIS] The pilot was certificated, current, and qualified in accordance with federal regulations and company requirements to conduct the VFR flight. There was no evidence the pilot had any preexisting medical condition, and postmortem toxicology testing revealed negative results for alcohol and other tested-for substances, such as various potentially impairing prescription, over-the-counter, and illicit drugs. A review of information about the pilot’s recent activities revealed adequate sleep opportunities in the days before the accident and no evidence of acute or chronic sleep loss or circadian disruption. • Helicopter malfunction or failure. Examinations of the helicopter’s structures, engines, systems, and instruments identified no evidence of preimpact malfunction or failure that would have precluded normal operation. • Pressure on the pilot from the company, air charter broker, or client to complete the flight. Island Express’ policy (as specified in the GOM) stated that, if weather conditions began to deteriorate such that a pilot could not maintain the company minimum flight altitude and visibility (300 ft agl and 1 nm, respectively) flight operations will be stopped and not resume until the weather has improved. Minutes from company safety meetings showed repeated company support for these policies, including the DO’s emphasis that pilots should divert and land when faced with adverse weather, even if it meant transporting the passengers by taxi and getting a hotel room. Both the DO and the safety officer expressed confidence in the accident pilot’s judgment and ability to make sound weather-related decisions. Regarding the accident client, in the event that a pilot needed to decline accepting, cancel, or delay a flight, the pilot or other company personnel would notify the air charter broker, and the air charter broker would notify the client (or his representative). There was no evidence to suggest that the air charter broker or the client (or any client representative) ever challenged any Island Express pilot’s decision to delay, cancel, or terminate a flight. Thus, the NTSB concludes that none of the following safety issues were identified for the accident flight: (1) pilot qualification deficiencies or impairment due to medical condition, alcohol, other drugs, or fatigue; (2) helicopter malfunction or failure; or (3) pressure on the pilot from Island Express, the air charter broker, or the client to complete the flight. The ATC services provided by the SNA and VNY tower controllers and the BUR tower controller before the flight entered the BUR class C airspace were unremarkable and in accordance with FAA procedures. Although the BUR tower controller did not inform the pilot that radar NTSB Aircraft Accident Report 34 contact was established with the flight, this procedural deviation did not affect the flight’s transition through the BUR airspace.

[ANALYSIS] However, the cloud layer would have been more optically thick from about 2,000 ft msl to the top of the marine layer. Thus, the NTSB concludes that, at the time that the pilot took action to initiate a climb, the helicopter had already begun penetrating clouds, and the pilot lost visual reference to the horizon and the ground. The loss of outside visual reference was possibly intermittent at first but likely complete by the time the flight began to enter the left turn that diverged from its route over US 101. 2.2.2.2 Pilot’s Inadequate Adverse Weather Avoidance Although the regional weather conditions that the accident flight encountered were atypical for January, there was no evidence that any dynamic or rapidly deteriorating weather phenomena NTSB Aircraft Accident Report 37 were present that would have prevented the pilot from discontinuing the flight or changing his planned route and maneuvering the helicopter to remain in VMC. According to FAA guidance, when approaching IMC, slowing the helicopter can reduce the closure rate between the helicopter and the adverse weather conditions (FAA 2019e, 11-25). This can allow a pilot more time to safely maneuver the helicopter to avoid the conditions. A EuroSafety instructor who provided supplemental training to the accident pilot taught this procedure, and one company pilot said that airspeeds as low as 60 kts would be appropriate, depending on the weather conditions and the terrain, to allow time to maneuver. According to the POI who evaluated the accident pilot during checkrides, appropriate adverse-weather-avoidance maneuvers may include diverting, returning to base, or landing the helicopter. The helicopter performance study determined that, while the flight was over US 101, the helicopter’s groundspeed remained about 140 kts. The helicopter’s flight path did not diverge from overflying US 101 until after the pilot had penetrated clouds and climbed into what was likely solid IMC. Although the pilot’s adverse-weather-avoidance training emphasized avoiding entry into IMC by slowing the helicopter and maneuvering or landing, there was no evidence that he attempted to do so. Thus, the NTSB concludes that the pilot’s poor decision to fly at an excessive airspeed for the weather conditions was inconsistent with his adverse-weather-avoidance training and reduced the time available for him to choose an alternative course of action to avoid entering IMC. 2.2.3 Pilot’s Spatial Disorientation As described previously, the pilot’s 0944:34 announcement to the SCT controller that he was initiating a climb occurred about 2 seconds after the helicopter began penetrating clouds. According to the helicopter performance study, the helicopter immediately began climbing at a rate of about 1,500 fpm and began to bank to the left. The ADS-B data showed that, at first, the helicopter’s flight path turned gradually left, generally continuing to follow US 101 below.

[ANALYSIS] Although the DO stated that he expected pilots to complete a flight risk analysis form before a flight and any time there was a significant change in weather, this expectation was not contained in any company guidance, and the accident pilot did not complete a new form before departing from SNA. (See section 2.4.1 for further discussion about company oversight of the flight risk analysis forms.) Per the flight risk analysis form, visibility below 3 miles (a nine-point risk item) would require the pilot to call the DO to discuss the weather, and en route SVFR (an eight-point risk item) would require the pilot to list an alternative plan for the flight. Thus, the NTSB concludes that, had the pilot completed an updated flight risk analysis form for the accident flight that considered the weather information available at the time the flight departed, the flight would have remained within the company’s low-risk category but would have required the pilot to seek input from the DO and to provide an alternative plan. The role of a lack of an alternative plan in the pilot’s decision-making process is discussed in section 2.3. 2.2.2 Flight’s Entry into Instrument Meteorological Conditions 2.2.2.1 Regional Weather Conditions and Localized Phenomena The pilot had flown the same passengers to the same destination the previous morning, having been able to fly a more direct route from SNA to CMA and at a higher altitude. However, on the day of the accident, an unusually thick marine layer was present across the region, such that NTSB Aircraft Accident Report 36 clouds associated with it extended up to 2,400 to 2,500 ft msl. Further, near-surface relative humidity near the accident site was still 100% at 1000, whereas it usually decreased earlier. In the type of low-level weather environment that existed that day, the cloud bases may not have been distinct (that is, the locations of the bases were not easy to visually identify). Also, the altitudes of the cloud bases likely would have varied across the area. As described previously, the flight’s departure from SNA and its progress through the BUR and VNY airspace en route to CMA were uneventful. At 0942:45 (about 3 minutes before the accident), the helicopter reached US 101 and began to follow it west toward CMA at an altitude of about 1,400 ft msl (550 ft agl) and groundspeed of about 140 kts.

[ANALYSIS] In addition, before the accident flight departed, the pilot had discussed with the air charter broker’s owner his plan to fly the route around BUR and VNY before joining US 101 to fly west to CMA. However, as described in section 2.2.1, the pilot did not complete an updated flight risk analysis form, which would have required him to list an alternative plan based on the weather information available at the time of departure. Devising an alternative plan before departure would have aided the accident pilot’s decision-making in flight concerning whether to continue the flight as originally planned or divert and land at an alternate destination. The pilot’s continuation of the accident flight into IMC was inconsistent with his typical judgment and decision-making behavior observed by the DO, safety officer, and company pilots interviewed.60 However, the pilot may have experienced plan continuation bias, which is an unconscious cognitive bias to continue with the original plan despite changing conditions (Woods 2020, 10). At the time that the flight began entering IMC, it was only about 25 miles from CMA (the destination), which had been reporting weather conditions above the basic VFR minimums since before the accident flight departed. With plan continuation bias, the closer the pilot gets to the destination, the stronger the bias becomes (Woods 2020, 10). Plan continuation bias (also often referred to as “get-there-itis” or plan continuation error) is known to negatively affect aeronautical decision-making, and it is addressed in various FAA guidance, training, and testing materials for pilots. Thus, the NTSB concludes that the pilot’s decision to continue the flight into deteriorating weather conditions was likely influenced by his self-induced pressure to fulfill the client’s travel needs, his lack of an alternative plan, and his plan continuation bias, which strengthened as the flight neared the destination. 60 The one company pilot who said he heard negative “stories” about the accident pilot’s decision-making did not personally witness any risk-taking behavior. NTSB Aircraft Accident Report 41 2.4 Incomplete Implementation of Safety Management System 2.4.1 Company Oversight of Flight Risk Analysis Forms Island Express had an SMS that was neither required by the FAA nor part of the company’s FAA-approved or -accepted programs. As such, the SMS did not receive (and was not required to receive) any FAA oversight. Although the provisions in the SMS Manual stated that the DO was the accountable executive who reported to the company president, there was no evidence that the president was actively involved with the SMS or mandated any company compliance with it. The company used some SMS risk management tools but did not implement the entire SMS as outlined in its SMS Manual. One SMS tool that company pilots used was the flight risk analysis form. The flight risk analysis forms were intended to document the risks associated with each flight, provide specific mitigations for certain risk items, and ensure that company management evaluated any planned flight that met the elevated- or high-risk criteria.

[CONCLUSIONS > FINDINGS] However, we note that fiscal year 2020 has ended, and the action has not been taken. Therefore, we are classifying Safety Recommendation A-13-13 “Open—Unacceptable Response.” NTSB Aircraft Accident Report 52 3. Conclusions 3.1 Findings 1. None of the following safety issues were identified for the accident flight: (1) pilot qualification deficiencies or impairment due to medical condition, alcohol, other drugs, or fatigue; (2) helicopter malfunction or failure; or (3) pressure on the pilot from Island Express Helicopters Inc., the air charter broker, or the client to complete the flight. 2. Although the air traffic controller’s failure to report the loss of radar contact and radio communication with the accident flight was inconsistent with air traffic control procedures, this deficiency did not contribute to the accident or affect its survivability. 3. Had the pilot completed an updated flight risk analysis form for the accident flight that considered the weather information available at the time the flight departed, the flight would have remained within the company’s low-risk category but would have required the pilot to seek input from the director of operations and to provide an alternative plan. 4. At the time that the pilot took action to initiate a climb, the helicopter had already begun penetrating clouds, and the pilot lost visual reference to the horizon and the ground. The loss of outside visual reference was possibly intermittent at first but likely complete by the time the flight began to enter the left turn that diverged from its route over US Route 101. 5. The pilot’s poor decision to fly at an excessive airspeed for the weather conditions was inconsistent with his adverse-weather-avoidance training and reduced the time available for him to choose an alternative course of action to avoid entering instrument meteorological conditions. 6. The pilot experienced spatial disorientation while climbing the helicopter in instrument meteorological conditions, which led to his loss of helicopter control and the resulting collision with terrain. 7. The pilot’s decision to continue the flight into deteriorating weather conditions was likely influenced by his self-induced pressure to fulfill the client’s travel needs, his lack of an alternative plan, and his plan continuation bias, which strengthened as the flight neared the destination. 8. Island Express Helicopters Inc.’s lack of a documented policy and safety assurance evaluations to ensure that its pilots were consistently and correctly completing the flight risk analysis forms hindered the effectiveness of the form as a risk management tool. 9. A fully implemented, mandatory safety management system could enhance Island Express Helicopters Inc.’s ability to manage risks. 10. The use of appropriate simulation devices in scenario-based helicopter pilot training has the potential to improve pilots’ abilities to accurately assess weather and make appropriate weather-related decisions. NTSB Aircraft Accident Report 53 11. Objective research to evaluate spatial disorientation simulation technologies may help determine which applications are most effective for training pilots to recognize the onset of spatial disorientation and successfully mitigate it. 12.

[FINDINGS] The loss of outside visual reference was possibly intermittent at first but likely complete by the time the flight began to enter the left turn that diverged from its route over US Route 101. • The pilot’s poor decision to fly at an excessive airspeed for the weather conditions was inconsistent with his adverse-weather-avoidance training and reduced the time available for him to choose an alternative course of action to avoid entering instrument meteorological conditions. NTSB Aircraft Accident Report ix • The pilot experienced spatial disorientation while climbing the helicopter in instrument meteorological conditions, which led to his loss of helicopter control and the resulting collision with terrain. • The pilot’s decision to continue the flight into deteriorating weather conditions was likely influenced by his self-induced pressure to fulfill the client’s travel needs, his lack of an alternative plan, and his plan continuation bias, which strengthened as the flight neared the destination. • Island Express Helicopters Inc.’s lack of a documented policy and safety assurance evaluations to ensure that its pilots were consistently and correctly completing the flight risk analysis forms hindered the effectiveness of the form as a risk management tool. • A fully implemented, mandatory safety management system could enhance Island Express Helicopters Inc.’s ability to manage risks. • The use of appropriate simulation devices in scenario-based helicopter pilot training has the potential to improve pilots’ abilities to accurately assess weather and make appropriate weather-related decisions. • Objective research to evaluate spatial disorientation simulation technologies may help determine which applications are most effective for training pilots to recognize the onset of spatial disorientation and successfully mitigate it. • A flight data monitoring program, which can enable an operator to identify and mitigate factors that may influence deviations from established norms and procedures, can be particularly beneficial for operators like Island Express Helicopters Inc. that conduct single-pilot operations and have little opportunity to directly observe their pilots in the operational environment. • A crash-resistant flight recorder system that records parametric data and cockpit audio and images with a view of the cockpit environment to include as much of the outside view as possible could have provided valuable information about the visual cues associated with the adverse weather and the pilot’s focus of attention in the cockpit following the flight’s entry into instrument meteorological conditions.

[ANALYSIS] Thus, the NTSB concludes that the pilot experienced spatial disorientation while climbing the helicopter in IMC, which led to his loss of helicopter control and the resulting collision with terrain. 57 Although the pilot did not declare an emergency, the controller’s actions to identify and provide service to the accident flight were nearly identical to the initial actions specified for responding to an emergency. ATC procedures for responding to a pilot-declared emergency specify, in part, that the controller obtain the minimum information required to provide assistance, including the aircraft identification and type, nature of the emergency, and the pilot’s desires (FAA 2019b, 10-2-1). These minimum information criteria are nearly identical to the information the controller requested from the accident pilot. 58 Even though the pilot was familiar with the flight deck layout, to perform the task of pushing the IDENT button, he needed to direct his gaze, likely accompanied by head movement, to the center panel where the button was located. He also needed to remove his left hand from the helicopter’s collective control to reach toward the center panel to push the IDENT button while his right hand remained on the cyclic control. 59 When an aircraft accelerates, the pilot’s vestibular system will detect the linear acceleration and lead to the incorrect perception that aircraft is climbing when it is not. Also, the forces the pilot experiences when the aircraft is in a turn can also give the illusion that the aircraft is climbing (SKYbrary 2020). NTSB Aircraft Accident Report 40 2.3 Influences on Pilot’s Decision to Continue Flight into Adverse Weather As discussed in section 2.1, there was no evidence to suggest that Island Express, the air charter broker, or the client placed any pressure on the accident pilot to accept the charter flight request or complete the flight in adverse weather. However, a number of factors were present that may have influenced the pilot to place pressure on himself (self-induced pressure) to complete the flight. For example, the accident pilot was the client’s preferred pilot, whom the client trusted to fly his children. Also, the air charter broker used Island Express exclusively for the accident client’s local helicopter transport needs because Island Express was the only operator that met the air charter broker’s standards for the client, and the client reportedly appreciated the relationship between the two companies. The accident pilot likely took pride in these positions of trust both with the client directly and within Island Express. Further, the pilot’s relationship with the client was friendly, and he likely did not want to disappoint the client by not completing the flight. In addition, before the accident flight departed, the pilot had discussed with the air charter broker’s owner his plan to fly the route around BUR and VNY before joining US 101 to fly west to CMA. However, as described in section 2.2.1, the pilot did not complete an updated flight risk analysis form, which would have required him to list an alternative plan based on the weather information available at the time of departure.

[ANALYSIS] According to the helicopter performance study, the helicopter immediately began climbing at a rate of about 1,500 fpm and began to bank to the left. The ADS-B data showed that, at first, the helicopter’s flight path turned gradually left, generally continuing to follow US 101 below. However, as the helicopter continued to climb into the cloud layer and IMC, the pilot’s associated loss of outside visual references would have required him to transition his focus to the helicopter’s flight instruments for reference to maintain awareness of the helicopter’s flight profile, including its pitch attitude, bank angle, and climb or descent rate. FAA guidance notes that the need to use outside visual references is natural for helicopter pilots and that avoiding entering IMC during a VFR flight is critical for even instrument-rated pilots in IFR-equipped helicopters. The guidance considers a VFR flight’s encounter with IMC, during which the pilot may be unprepared for the loss of visual reference, to be a life-threatening emergency (FAA 2019e, 11-24 and -25). This is because, following the loss of visual cues in flight, pilots are susceptible to experiencing vestibular illusions, which can lead to spatial disorientation and a loss of control of the aircraft.55 Vestibular illusions occur when the human vestibular system of the inner ear produces a false sense of helicopter attitude and trajectory. The vestibular system allows a person to have a 55 One study of US Army helicopter operations found that sudden loss of visual cues preceded about 25% of helicopter accidents attributed to spatial disorientation (Braithwaite et al. 1998, 1031-7). NTSB Aircraft Accident Report 38 sense of balance and spatial orientation. However, the vestibular system cannot distinguish between accelerations and tilt. Additional sensory inputs, such as visual cues, are needed for a person to correctly perceive attitude, bank angle, and acceleration. In the absence of outside visual references, a pilot’s misperception of any of these flight conditions can result in spatial disorientation. A pilot’s consistent scan and correct interpretation of the flight instruments and belief in their representation while operating in IMC can enable the pilot to resist reacting to compelling vestibular illusions and prevent spatial disorientation. The accident pilot was trained that, to recover from entry into IMC, he should first adjust the helicopter’s pitch and power to establish a stabilized, positive rate of climb at an airspeed of about 75 to 80 kts, then transition to using the autopilot before communicating with ATC. FAA guidance stated, in part, the following: Once the helicopter is stabilized, the pilot should declare an emergency with [ATC]. It is imperative that the pilot commit to controlling the helicopter and remember to aviate, navigate, and finally communicate. Often communication is attempted first, as it is natural to look for help in stressful situations.

[ANALYSIS] Weather conditions reported to the pilot by air traffic controllers at BUR and VNY during the flight included an overcast ceiling at 1,100 ft agl and visibility of 2.5 miles with haze. At 0944:34 (about 2 minutes before the accident), while the helicopter was flying west at an altitude of about 1,370 ft msl (450 ft agl) over US 101 and rising terrain, the pilot announced to an ATC facility that he was initiating a climb to get the helicopter “above the [cloud] layers,” and the helicopter immediately began climbing at a rate of about 1,500 fpm. About the same time, the helicopter began a gradual left turn, and its flight path generally continued to follow US 101 below. About 36 seconds later and while still climbing, the helicopter began to turn more tightly to the left, and its flight path diverged from its overflight of US 101. The helicopter reached an altitude of about 2,370 ft msl (about 1,600 ft agl) at 0945:15, then it began to descend rapidly in a left turn to the ground. At 0945:17 (while the helicopter was descending), the air traffic controller asked the pilot to “say intentions,” and the pilot replied that the flight was climbing to 4,000 ft msl. A witness near the accident site first heard the helicopter then saw it emerge from the bottom of the cloud layer in a left-banked descent about 1 or 2 seconds before impact. The following analysis discusses the accident sequence and evaluates the following safety issues: • The pilot’s preflight weather and flight risk planning (section 2.2.1); • The flight’s entry into IMC (section 2.2.2), and the pilot’s inadequate adverse weather avoidance (section 2.2.2.2); • The pilot’s spatial disorientation (section 2.2.3); • Factors influencing the pilot’s decision to continue the flight into adverse weather, including his self-induced pressure, lack of an alternate plan, and plan continuation bias (section 2.3); • Island Express’ incomplete implementation of its SMS (section 2.4), including its lack of safety assurance evaluations, such as those that could have ensured the effectiveness of the flight risk analysis forms (section 2.4.1); • The benefits of a mandatory SMS (section 2.4.2); NTSB Aircraft Accident Report 33 • The benefits of flight simulation devices for pilot training in adverse weather avoidance (section 2.5); • The benefits of a flight data monitoring (FDM) program (section 2.6); and • The value of crash-resistant flight recorder systems in preventing future accidents (section 2.7). Having completed a comprehensive review of the circumstances that led to the accident, the investigation identified none of the following safety issues for the accident flight: • Pilot qualification deficiencies or impairment due to medical condition, alcohol, other drugs, or fatigue.

[CONCLUSIONS > FINDINGS] The use of appropriate simulation devices in scenario-based helicopter pilot training has the potential to improve pilots’ abilities to accurately assess weather and make appropriate weather-related decisions. NTSB Aircraft Accident Report 53 11. Objective research to evaluate spatial disorientation simulation technologies may help determine which applications are most effective for training pilots to recognize the onset of spatial disorientation and successfully mitigate it. 12. A flight data monitoring program, which can enable an operator to identify and mitigate factors that may influence deviations from established norms and procedures, can be particularly beneficial for operators like Island Express Helicopters Inc. that conduct single-pilot operations and have little opportunity to directly observe their pilots in the operational environment. 13. A crash-resistant flight recorder system that records parametric data and cockpit audio and images with a view of the cockpit environment to include as much of the outside view as possible could have provided valuable information about the visual cues associated with the adverse weather and the pilot’s focus of attention in the cockpit following the flight’s entry into instrument meteorological conditions. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the pilot’s decision to continue flight under visual flight rules into instrument meteorological conditions, which resulted in the pilot’s spatial disorientation and loss of control. Contributing to the accident was the pilot’s likely self-induced pressure and the pilot’s plan continuation bias, which adversely affected his decision-making, and Island Express Helicopters Inc.’s inadequate review and oversight of its safety management processes. NTSB Aircraft Accident Report 54 4. Safety Recommendations 4.1 New Recommendations As a result of this investigation, the National Transportation Safety Board makes the following new safety recommendations. To the Federal Aviation Administration: Require the use of appropriate simulation devices during initial and recurrent pilot training for Title 14 Code of Federal Regulations Part 135 helicopter operations to provide scenario-based training that addresses the decision-making, skills, and procedures needed to recognize and respond to changing weather conditions in flight, identify and apply mitigation strategies for avoiding adverse weather, practice the transition to the use of flight instruments to reduce the risk of spatial disorientation, and maintain awareness of a variety of influences that can adversely affect pilot decision-making. (A-21-5) Convene a multidisciplinary panel of aircraft performance, human factors, and aircraft operations specialists to evaluate spatial disorientation simulation technologies to determine which applications are most effective for training pilots to recognize the onset of spatial disorientation and successfully mitigate it, and make public a report on the committee’s findings. (A-21-6) To Island Express Helicopters Inc.: Participate in the Federal Aviation Administration’s Safety Management System Voluntary Program. (A-21-7) Install flight data recording devices capable of supporting a flight data monitoring (FDM) program on each helicopter in your fleet and establish an FDM program that reviews all available data sources to identify deviations from established norms and procedures as well as other potential safety issues. (A-21-8) 4.2 Previously Issued Recommendations Reiterated in This Report The National Transportation Safety Board reiterates the following safety recommendations.

[ANALYSIS] The ability to discern these objects and terrain is the seeing condition, and is related to the amount of natural and man-made lighting available, and the contrast, reflectivity, and texture of surface terrain and obstruction features. In order to conduct operations safely, seeing conditions must be accounted for in the planning and execution of night VFR operations.” NTSB Aircraft Accident Report 51 view, reduced image resolution, and the presence of digital noise. Low lighting conditions can result in a lower contrast NVG image and increased digital noise. Such images are more difficult to interpret and may cause a tendency to fly lower in an effort to improve image quality. The presence of meteorological obscurants like rain or snow has the potential to further degrade NVG image quality. The effect of precipitation on image quality can be unpredictable and can change with the nature and intensity of the precipitation. Thus, meteorological and astronomical forecasts that included low light, rain, and snow indicated the potential for degraded NVG effectiveness and increased risk of an inadvertent encounter with IMC. Therefore, the NTSB concludes that, at the time the pilot was notified about the stranded snowmobiler, sufficient information was available to indicate that the mission carried a high degree of risk due to the weather and low lighting conditions. The investigation revealed no evidence that Alaska DPS managers ever pressured the pilot to accept or complete a flight. Thus, it does not appear that the pilot was subjected to any direct management pressure to accept or continue SAR missions. However, the pilot was described as having exceptionally high motivation for flying-related tasks, and he took great pains to make sure that he and the helicopter were always available for any DPS missions. He had frequent conflicts with maintenance personnel over the timeliness of required maintenance and rarely took time off because he did not want to miss opportunities for flying. The pilot reportedly enjoyed flying the helicopter and had achieved a high level of VFR helicopter flying proficiency. Putting this skill to use likely provided some intrinsic satisfaction. The pilot’s spouse said that the pilot was very close to his own family, and he appreciated being able to bring other people safely back to theirs. In addition, records indicate that the pilot had received a great deal of public recognition for past rescues. His personnel file contained many heartfelt letters of thanks from people he had rescued, and he had received several high-profile awards. In addition, a substantial amount of the pilot’s income came from being on call and flying missions outside of his scheduled work hours. Colleagues and supervisors said that the pilot was very sensitive to any changes in aircraft section operating policies that could reduce his pay, such as reducing his standby time by using the relief pilot. The relief pilot said that the pilot feared being replaced if other pilots were allowed to fly more missions. As a result of these multiple sources of motivation, the pilot carefully guarded his role as the helicopter’s primary pilot.

[ANALYSIS] The success of these past missions, particularly those involving poor weather, likely increased the pilot’s confidence that he could safely continue VFR flight at night in marginal weather conditions. A precautionary landing would have stranded the pilot, trooper, and hypothermic snowmobiler in an uncomfortable (but probably survivable) situation until the weather improved NTSB Aircraft Accident Report 54 or ground resources could assist them.58 Executing a landing in the dark, reduced-visibility conditions in an unfamiliar clearing with heavy snow on the ground might also damage the helicopter. Continuing VFR flight, on the other hand, increased the pilot’s risk of experiencing a weather-related accident, but the risk of this type of accident probably seemed remote to the pilot, given his past experience. Thus, the pilot had to choose between two undesirable alternatives: one that involved a high perceived likelihood of inconvenience and possible helicopter damage and another that involved a low perceived likelihood of a serious accident. The pilot chose the latter option, and the risk of a serious accident was realized. The NTSB concludes that the pilot’s exceptionally high motivation for SAR missions and past successes likely increased his risk tolerance and influenced his decision to continue flying in deteriorating weather conditions and risk a weather-related accident rather than accept the certain inconveniences and potential hazards associated with a precautionary landing. 2.4 Organizational Issues 2.4.1 Risk Assessment The accident helicopter was a single-engine, nonIFR-certified platform and was crewed by a single pilot who was not instrument-current and had NVGs. This meant that the equipment/crew pairing was capable of operating in dark night VMC conditions but not in IMC. Inadvertent encounters with IMC would result in a high risk. One prudent organizational strategy for managing this risk should have entailed establishing minimum VFR weather requirements that provide some degree of separation between the helicopter and weather conditions that could obscure a pilot’s view of the natural horizon, with or without NVGs. However, as discussed previously, the Alaska DPS did not apply across-the-board VFR weather minimums to its helicopter pilots, other than FAA requirements. The Part 91 regulations that applied to Alaska DPS flights required only that the helicopter be operated clear of clouds below 1,200 ft. In contrast, for HEMS operations conducted under Part 135, the FAA established NVG weather minimums. These minimums, which are part of a HEMS-specific operations specification, range from an 800-ft ceiling and 3-mi visibility for a local flight in nonmountainous terrain to a 1,000-ft ceiling and 5-mi visibility for cross-country flights in mountainous terrain.

[ANALYSIS] For example, a review of NTSB accident data shows that, since January 1, 2007, there were 466 accidents involving such aircraft, and these accidents claimed 246 lives. In addition, in 55 of these accidents, the probable cause statements contained some element of uncertainty, such as an undetermined cause or factor. 64 Safety Recommendation A-13-13 superseded Safety Recommendation A-03-64, which was issued on December 22, 2003, and asked the FAA to “[r]equire all turbine-powered, nonexperimental, nonrestricted-category aircraft that are manufactured prior to January 1, 2007, that are not equipped with a cockpit voice recorder, and that are operating under 14 [CFR] Parts 91, 135, and 121 to be retrofitted with a crash-protected image recording system by January 1, 2007.” That safety recommendation (which superseded Safety Recommendation A-99-60) was superseded by Safety Recommendation A-09-10 and, therefore, was classified “Closed—Unacceptable Action/Superseded” on February 9, 2009. NTSB Aircraft Accident Report 69 3. Conclusions 3.1 Findings 1. The pilot was qualified to fly search and rescue missions in visual meteorological conditions (but not instrument meteorological conditions) in the accident helicopter, and his performance was unlikely affected by medical factors, fatigue, or physical activities associated with the ground portion of the rescue activity. 2. The in-flight image recording and wreckage examinations showed that the helicopter and its engine were operating normally throughout the flight. No mechanical abnormalities with the helicopter were identified. 3. Soon after departure from the remote landing site, the helicopter likely encountered instrument meteorological conditions, which included low clouds, heavy snow, and near-zero-visibility conditions. 4. Although icing conditions were likely present during the accident flight, the performance of the helicopter does not appear to have been degraded at the time of the accident. 5. The pilot experienced a total loss of external visual references while operating in close proximity to terrain, which led him to attempt to transition to instrument flight. 6. The pilot’s action to cage the attitude indicator outside those conditions under which it could be safely caged indicates that he distrusted the information he was seeing. 7. The pilot’s caging of the attitude indicator made it very unlikely that he would regain control of the helicopter in instrument meteorological conditions. 8. The helicopter’s erratic maneuvers are consistent with the pilot’s spatial disorientation, a loss of control in flight, and his inability to recover the helicopter because of his lack of instrument experience and the lack of accurate attitude information. 9. When the pilot was contacted about the mission, forecasts indicated that conditions in the search area would be instrument flight rules and that forecast cloud ceilings and visibility would likely be below the pilot’s Alaska Department of Public Safety weather minimums and possibly below his last known personal weather minimums. 10.

[ANALYSIS] NTSB Aircraft Accident Report 53 2.3.3 Decision to Continue Mission Before the accident flight, when the pilot first arrived in the search area about 2200, he flew the helicopter between 1,100 and 1,200 ft msl, which suggests that the cloud ceiling was at least 650 to 750 ft agl about the time that he landed at the remote rescue location. However, the altitude that the pilot initially flew during the accident flight was about 700 ft msl (250 ft agl), suggesting that the cloud ceiling and/or visibility in the area had deteriorated significantly during the time the helicopter was on the ground. In addition, weather information and witness reports indicate a strong possibility of icing. The safest course of action at this point was to perform a precautionary landing. An examination of the terrain along the helicopter’s ground track identified several open areas that could have served as emergency landing areas for the helicopter. However, due to the changing precipitation and low lighting conditions, it is uncertain whether the pilot could see these potential landing areas well enough to determine whether they were suitable for landing. Although it is possible that the condition of the snowmobiler created a sense of urgency that prompted the pilot to push on in deteriorating conditions, there is insufficient evidence about the seriousness of the snowmobiler’s medical condition to know how it might have been perceived by the pilot and flight observer. Although the flight observer reported to a dispatcher that the snowmobiler was hypothermic, he did not communicate any detailed information about the snowmobiler’s condition, and hypothermia cases can range from mild to severe. A factor that likely influenced the pilot’s continued VFR flight in deteriorating weather was his high motivation for performing missions and accomplishing rescues. Although the pilot was described by colleagues as being very safety oriented, the aircraft section commander had expressed concern to the pilot about the riskiness of some flights, and the pilot had responded that he agreed to do such things when he was hired and planned to continue doing them. As a result of his conversations with the pilot, the aircraft section commander said he believed the pilot appreciated the hazards associated with risky decisions but that he felt a self-imposed obligation to take certain risks to accomplish rescues. Another factor that likely influenced the pilot’s continued VFR flight into deteriorating weather was an increased tolerance for risk as a result of successful past outcomes. Although the pilot had experienced a takeoff accident 7 years earlier involving white-out conditions and loss of visibility from snowfall and snow on the ground that billowed up in the rotor wash, he had not experienced any other accidents since, despite conducting additional missions that often involved high-risk activities, such as maneuvering through areas of poor weather at night and flying inches above fast-moving bore tides. The success of these past missions, particularly those involving poor weather, likely increased the pilot’s confidence that he could safely continue VFR flight at night in marginal weather conditions.

[ANALYSIS] Create a formal TFO training program that includes training on aeronautical decision-making, crew resource management, and operating aircraft navigational and communications equipment, and use TFOs during SAR operations. Develop and implement a comprehensive SMS for aircraft operations that (1) holds senior state personnel accountable for the safety of state law enforcement aircraft operations, (2) is tailored to the department’s missions, and (3) is based on industry best practices. Arrange for an audit of the SMS implemented in response to Safety Recommendation A-14-105 to be conducted every 3 years by an outside organization. 2.6 Attitude Indicator Limitations As discussed in section 2.2, about 40 seconds after the helicopter entered IMC, the pilot caged the attitude indicator, likely because he distrusted the information the instrument was displaying. Although the reason the pilot distrusted the information cannot be known, the investigation considered two possible explanations. One possible explanation is that the pilot might have distrusted the attitude display because he was spatially disoriented. Maneuvering flight without external visual references can lead to a variety of illusions of motion, which can result in inaccurate perceptions of an aircraft’s attitude and trajectory. A number of risk factors for spatial disorientation preceded the pilot’s operation of the caging knob. These included the pilot’s lack of instrument flying currency, the loss of external visual references, his unplanned transition to instrument flight, aggressive maneuvering, and operational distractions related to setting up the navigational instruments for NTSB Aircraft Accident Report 65 flight in IMC. Research indicates that spatial disorientation can result in false perceptions of instrument malfunction. In a 2002 spatial disorientation survey, for example, 18% of US Air Force pilots operating rotary wing aircraft reported having experienced at least one instance of illusory instrument malfunction (Matthews and others 2002). Another possible explanation is related to the instrument’s limitations. According to information provided by the attitude indicator’s manufacturer, the AIM 1200 attitude indicator is limited to indicating ± 25° of pitch. Thus, if an aircraft were to operate at a pitch that exceeded the limitation, the pitch indicator would stop and remain at the limit until the pitch no longer exceeded the limitation. Image evidence shows that, during the first 30 seconds after the helicopter entered IMC, the pitch increased from about 0° to at least 17° nose up. Although pitch indications on the attitude indicator higher than about 17° could not be accurately measured from the cockpit images, the images show that the indicated pitch remained above 17° from 2318:28 to 2318:40. This is consistent with the attitude indicator stopping at 25° and remaining there as the helicopter continued pitching up. Although the operating manual for the AIM 1200 did not include information about the pitch indicating range limits, even if it had, and the pilot were aware of it, it is uncertain whether the pilot would have immediately understood this instrument behavior upon encountering it in a high-stress, high-workload situation.

[ANALYSIS] During instrument helicopter training, which the pilot had completed many years earlier, he was trained in partial-panel techniques, including using the turn-and-bank indicator as a secondary source of obtaining information about the helicopter’s bank attitude. However, the turn-and-bank indicator was inoperative during the accident mission because the pilot had previously disabled it. Thus, this source of bank information was not available to help the pilot determine the helicopter’s attitude as he tried to maintain turn-and-bank control. The absence of a functioning turn-and-bank indicator might have been moot because the pilot had minimal (0.5 hour) helicopter actual instrument flying experience, lacked helicopter instrument flying currency, and had no recent instrument training. Therefore, it is unlikely that he would have been able to maintain control of the helicopter using partial-panel techniques during the climbing turn, even with a working turn-and-bank indicator. Research involving instrument-rated, fixed-wing pilots suggests that maintaining aircraft control following a simulated attitude instrument failure in actual instrument conditions with a working turn-and-bank indicator is extremely difficult and leads to loss of control in at least 10% of cases (Roy and Beringer 2002). The success rate for a helicopter pilot during aggressive, low-speed maneuvering in a nonIFR-certificated helicopter would likely be much lower, due to the inherently unstable nature of such helicopters (compared to IFR-equipped helicopters) and the even greater dependence of their pilots on external visual cues for maintaining helicopter control. Therefore, the NTSB concludes that the pilot’s caging of the attitude indicator made it very unlikely that he would regain control of the helicopter in IMC. The NTSB further concludes that the helicopter’s erratic maneuvers are consistent with the pilot’s spatial disorientation, a loss of control in flight, and his inability to recover the helicopter because of his lack of instrument experience and the lack of accurate attitude information. NTSB Aircraft Accident Report 50 2.3 Pilot’s Risk Management Considerations 2.3.1 Decision to Accept Mission The pilot did not call a flight service specialist for a weather briefing, and the investigation was unable to determine which weather information sources the pilot may have examined before deciding to accept the mission. Based on the available weather information products and the standard services provided by a qualified weather briefer, it is likely that, had the pilot called for a briefing at the time that he was notified of the mission, the briefer would have informed the pilot about the radar-depicted line of light-to-moderate echoes that was moving toward TKA. A review of the typical sources used by Alaska DPS pilots revealed that the FA forecasted visibilities as low as 4 mi in places that included the search area with isolated rain and snow showers, and the TKA TAF issued at 2008 forecasted a cloud ceiling of 1,000 ft agl at the airport.

[ANALYSIS] Review of ADW weather data showed that the weather given to the pilot had been issued at 1855 local time, almost 5 hours earlier. Weather information available to controllers is time-stamped, and the controller should have noticed that the report was outdated. The controller did not appear to have noticed the time discrepancy and did not advise the pilot of the age of the report or take any action to contact ADW for current weather information. During the time that the aircraft was being vectored for the approach, ADW was reporting wind from 110° at 3 knots, visibility 7 miles, cloud ceiling 1,300 broken, temperature 20° C, dew point 20 °C. Although these weather conditions were well above the localizer-only and ILS landing minimums for the runway 19R approaches, the cloud ceiling was 500 feet lower than that reported to Trooper 2. If the pilot had been given the current weather information, he would have expected to be unable to see the ground until he had reached a significantly lower altitude. Knowing that the ceiling was 500 feet lower may have discouraged the pilot from attempting to duck under the cloud ceiling and resulted in his continuing the approach to the airport at a 500 fpm rate of descent. Therefore, the NTSB concludes that the failure of the PCT controller to provide the current ADW weather information likely led the pilot to expect that he could descend below the cloud ceiling and establish visual contact with the ground at an altitude well above the MDA for the approach. The controller did not explicitly issue Trooper 2 an instrument clearance or an IFR transponder code. An IFR transponder code is necessary for aircraft to receive MSAW service. When interviewed, he stated that the pilot never requested an instrument clearance, and the controller believed that the pilot wanted a VFR practice approach to ADW. As the pilot had requested IFR from DCA and there was no further discussion of VFR operations by either the pilot or the PCT controller after he was handed off, the pilot could have reasonably expected that he was operating under IFR despite not having received an explicit IFR clearance from PCT. The NTSB Aircraft Accident Report 57 controller's statement that he believed Trooper 2 was VFR all the way to touchdown is contravened by the controller's failure to restrict the aircraft to VFR operations as required by FAA directives. Between the time Trooper 2 contacted PCT and issuance of the IFR approach clearance, Trooper 2‘s status as an IFR or VFR flight was ambiguous. Once the helicopter was cleared for an approach without a VFR restriction, Trooper 2 was an IFR flight and should have been given an IFR transponder code.

[ANALYSIS] The recovered components showed no evidence of any preimpact structural, engine, or system failures. IMC prevailed in the accident area with scattered clouds near 200 feet agl and cloud ceilings near 500 feet agl. Although ADW was reporting visibility of 4 miles in mist, lower visibilities in fog occurred locally in the accident area. This analysis will address the following issues: risk assessments, pilot performance and training, TAWS, ATC deficiencies, SYSCOM duty officer performance, and emergency response. Patient transport decisions, flight recorder requirements, and oversight are also discussed. 2.2 Pilot’s Decision to Accept the Flight At the beginning of his shift, the pilot obtained a computerized weather briefing from the FAA‘s DUAT service. Although the briefing the pilot obtained included observations, forecasts, and other information, it did not include weather hazards, a topic the pilot chose not to include in the briefing. If the pilot had included weather hazards, the briefing would have contained an AIRMET for IFR conditions issued at 1645 and valid until 2300 for an area that began immediately north and east of ADW, covered eastern Maryland and Delaware, and extended through New England. PGH was located within the boundaries of the AIRMET; ADW and the landing zone at Waldorf were not. At the time of the briefing, the current weather at ADW indicated VFR conditions with a temperature/dew point spread of 1° C. The ADW forecast weather until 0100 on September 28 indicated VFR conditions, and between 0100 and 0200, the weather at ADW was expected to deteriorate to IFR conditions with the ceiling broken at 500 feet and overcast at 1,000 feet. The DCA forecast indicated that by 2200, the weather conditions were expected to deteriorate to IFR conditions with scattered clouds at 400 feet and ceiling overcast at 800 feet. These conditions were above the MSP Aviation Command minimums for acceptance of a night medevac flight (800 foot ceiling, 3 miles visibility). However, the ADW and DCA forecasts of deteriorating NTSB Aircraft Accident Report 49 weather and the small temperature/dew point spread at ADW should have alerted the pilot to the possibility of local fog or low cloud formation as the evening progressed. According to FAA AC 00-6A, ―Aviation Weather,‖ pilots should anticipate fog when the temperature-dew point spread is 5° F (3.6° C) or less and decreasing. Since the weather conditions were close to MSP minimums and the forecast called for conditions to deteriorate, the section was accepting flights on a ―call by call‖ basis, and MSP procedures required the pilot to conduct a check of the weather every 2 hours. The procedures did not specify what weather sources were considered acceptable for use in obtaining the required weather updates. There is no record of the pilot receiving another weather briefing during his shift either on the computer from DUAT or by calling a weather briefer.

[ANALYSIS] This could have led the pilot to conclude the weather was better than reported. Regardless of the unavailability of the DoD surface weather observations, the pilot did have access to information that indicated the weather conditions were continuing to deteriorate throughout the evening. In addition to surface observations, the pilot could also have viewed terminal area forecasts and AIRMETs via the HEMS weather tool. He could have obtained the amended terminal forecast for DCA, issued at 1933, which indicated that weather conditions 71 Local (on airport) dissemination of the ADW observations was not affected by the DoD data communications failure. NTSB Aircraft Accident Report 50 were expected to deteriorate temporarily from VFR to IFR between 2000 and 2300 with the ceiling broken at 800 feet, overcast at 1,500 feet. Additionally, he could have obtained an AIRMET for IFR conditions issued at 2245 for an area that encompassed the entire route of Trooper 2. The forecast ceilings and visibilities in the DCA terminal forecast and in the AIRMET were above the MSP Aviation Command minimums for acceptance of a night medevac flight. However, the amendment to the DCA forecast and the issuance of the AIRMET indicated that weather conditions were continuing to deteriorate. Since HEMS weather tool data is not archived, it is not possible to determine what information was viewed by the pilot. It appears that the pilot based his decision to launch solely on the weather observations at College Park and DCA and the suitable conditions implied by the other medevac helicopter‘s completed flight. Other pertinent weather data—the low temperature/dew point spreads at ADW and College Park, the AIRMET for IFR conditions encompassing the route of flight, and the continuing deterioration of the weather conditions as the evening progressed—were either discounted by the pilot or not obtained. If the pilot had thoroughly obtained and reviewed all of the available weather information, it is likely he would have realized that there was a high probability of encountering weather conditions less than MSP minimums on the flight and this would have prompted him to decline the flight. Therefore, the NTSB concludes that the pilot‘s decision to accept the flight, after his inadequate assessment of the weather, contributed to the accident. According to the safety officer, at the time of the accident, MSP did not have a formal risk management program in place. He explained that there was optional guidance available to pilots in the form of a ―Risk Assessment Matrix.‖ However, review of the MSP Operations Manual revealed that it stated the flight crew ―will apply‖ the matrix and based on the risk assessment, increase visibility and ceiling minimums ―to the crew‘s comfort level prior to accepting the mission.‖ The matrix indicated that a temperature/dew point spread of less than 2° C raised the flight risk from low to medium risk. Although the matrix indicated that no flights were to be made if the risk level was high, it provided no instructions concerning medium-risk flights. There is no evidence indicating that the accident pilot consulted the matrix before the flight.

[ANALYSIS] The procedures did not specify what weather sources were considered acceptable for use in obtaining the required weather updates. There is no record of the pilot receiving another weather briefing during his shift either on the computer from DUAT or by calling a weather briefer. However, there is evidence that the pilot used the HEMS weather tool to check the weather before accepting the flight. Another pilot observed the HEMS weather tool displayed on the hangar computer after the accident, and the accident pilot‘s remarks suggest he was viewing the weather observations available through the tool during his conversation with the SYSCOM DO when he received the call for the flight. Normally, the surface weather observations for ADW, DCA, Fort Belvoir, and College Park are available through the HEMS weather tool. However, the weather reports for ADW and Fort Belvoir were not available on the night of the accident because of a technical failure of a DoD communications system (described in section 1.7.2). Therefore, the pilot was unaware that Fort Belvoir was reporting IFR conditions with a visibility of 1 1/4 miles in mist and a 0° temperature/dew point spread. It is unknown whether the pilot obtained the ADW weather (which was VFR conditions with a temperature/dew point spread of 0°) from another source, such as the airport‘s automatic terminal information service.71 During his conversation with the DO, the pilot specifically mentioned only the weather conditions at College Park and DCA. The weather reports for both of these locations met the MSP criteria for acceptance of a night medevac flight. However, College Park was at the 800-foot minimum ceiling for acceptance of a flight and was reporting a 0° temperature/dew point spread. The pilot‘s conversation with the DO indicated that the pilot was hesitant to accept the flight, as he was unsure he could make it to PGH due to deteriorating weather conditions. During the conversation, he stated that ―maybe [the first responders] would change their mind‖ about the flight request, indicating his reluctance to accept the flight. However, despite his misgivings, the pilot decided to accept the flight. The pilot‘s comments, ―if they can do it we can do it,‖ followed by, ―yeah we ought to be able to do it…we‘re going to try it,‖ show that his decision-making was influenced by the report that another medevac helicopter had just successfully completed a flight. This could have led the pilot to conclude the weather was better than reported. Regardless of the unavailability of the DoD surface weather observations, the pilot did have access to information that indicated the weather conditions were continuing to deteriorate throughout the evening. In addition to surface observations, the pilot could also have viewed terminal area forecasts and AIRMETs via the HEMS weather tool.

[ANALYSIS] Orasanu, L. Martin, and J. Davison, “Cognitive and Contextual Factors in Aviation Accidents,” in E. Salas and ciates, 2001), pp. 209-225.] NTSB Aircraft Accident Report 50 disorientation, which could lead to loss of control and/or a controlled flight into terrain accident. Had the pilot performed an interim risk assessment and considered the external circumstances or discussed them with the spotter,107 the NMSP dispatcher, or SAR ground personnel, he would have been more likely to recognize the potential hazards associated with an immediate takeoff and might have delayed his departure from the remote landing site until more favorable conditions prevailed. The NTSB concludes that the pilot exhibited poor decision-making when he chose to take off from a relatively secure landing site at night and attempt VFR flight in adverse weather conditions. 2.3 F re complex or perceptually demanding,108 experts on cold exposu actors Affecting the Pilot’s Decision-Making This section addresses factors that could have influenced the pilot’s decision-making both before accepting the mission and when he took off from the mountain to return to SAF, including fatigue, self-induced pressure, and situational stress. The NTSB considered whether environmental and/or physiological factors related to the cold temperature or the high altitude might have degraded the pilot’s decision-making in this case. The pilot spent about 50 minutes searching for the hiker and carrying her up a steep slope in very cold, windy conditions, in freezing precipitation, while dressed only in an unlined summer-weight flight suit and undergarments. Although cold stress can degrade cognitive performance, especially for tasks that a re at the U.S. Army Research Institute for Environmental Medicine indicated that, based on the pilot’s level of exertion and the terrain, the pilot’s metabolic rate would likely have been fairly high, offsetting the cold weather’s effects and minimizing the risk of hypothermia. After reviewing the circumstances of this accident for evidence of any physiological effects from the high altitude (the pilot was operating in the unpressurized helicopter for more than 2 hours before the accident), U.S. Army Research Institute for Environmental Medicine personnel indicated that the altitude would have affected the pilot very little as well. Because the pilot lived at an altitude of about 6,000 feet, he would likely have been sufficiently acclimatized to operate at higher altitudes.109 Since the time at altitude was relatively short (about 3 hours), there was likely little hypoxic effect on the pilot’s cognitive function. 107 The SAR commander spoke with the spotter while the pilot was retrieving the hiker and urged the spotter to remain in place and wait for ground teams to arrive if it was not safe to take off. However, it is not clear that the spotter shared this information with the pilot when he returned to the helicopter; the spotter did not recall the pilot raisi formance in Cold Environments,” in D.E. Lounsbury, R.F. ng the possibility of remaining on the mountain overnight. 108 R.G.

[ANALYSIS] Therefore, the NTSB recommends that the governor of the state of New Mexico revise or reinforce NMSP SAR policies to ensure direct communication between NMSP aviation units and SAR ground teams and field personnel during a SAR mission. 2.6 Instrument Flying According to the spotter, when the pilot departed the mountain for SAF, he pointed ou d to maneuver toward lower and more open terrain in VFR conditions. However, the low clouds and snow likely obscured the mountains and led to an inadvertent IMC encounter shortly after takeoff. Although the pilot had a fixed-wing instrument rating and met instrument currency and proficiency requirements for that rating, he did not have a helicopter instrument rating. The pilot’s lack of a helicopter instrument rating was not technically an issue when he accepted the accident mission because VMC prevailed and NMSP aviation section pilots were not required to have instrument ratings for helicopter operations. NMSP management personnel explained that e operations (for example, SAR missions) required VMC. Further, NMSP helicopter pilo e expected stay lear of uds. NTSB Aircraft Accident Report 61 As a result of its 2009 public hearing on HEMS safety and the investigative results of several 2008 HEMS accidents, the NTSB issued Safety Recommendation A-09-97, which recommended that public HEMS operators conduct scenario-based training, to include inadvertent flight into IMC, frequently enough to ensure proficiency. Although the accident flight was not a HEMS flight, if the accident pilot been trained in how to exit inadvertent IMC in a helico c operators and can affect the most pilots and operators, the NTSB recomm hat the m the accident helicopter’s 406-MHz ELT were primarily rs on areas near the accident site and for eventually locating both the survivor and th recomm pter, he might have followed different procedures (for example, he might have climbed to a safe altitude and contacted ATC for assistance) or a different route (with potentially better results) when the helicopter departed the landing site and thus avoided the subsequent collision with terrain. The NTSB concludes that, because the accident pilot did not have a helicopter instrument rating, experience in helicopter instrument operations, or training specific to inadvertent helicopter IMC encounters, he was not prepared to react appropriately to the loss of visual references that he encountered shortly after takeoff. Because ALEA has the broadest membership of publi ends that ALEA revise its accreditation standards to require that all pilots receive training in methods for safely exiting inadvertently encountered IMC for all aircraft categories in which they operate. 2.7 Emergency Locating Equipment The accident helicopter was equipped with an ELT that transmitted on both the 406- and 121.5-MHz frequencies. The ELT’s signal was not received by the two geostationary operational environmental satellites most likely because of the topography of the crash site and the relative positions of the two satellites. However, ELT signals were received by low-Earth polar orbiting satellites.

[ANALYSIS] By the time the pilot and hiker returned to the helicopter (about 50 minutes after the pilot left the helicopter to retrieve the hiker and more than 1 hour after sunset), the sleet had turned to snow, and the clouds had lowered. Other witnesses who were camping at a lower elevation nearb structural icing. In addition, the remote landing site was no longer visible. Yet the pilot quickly prepared the helicopter f tes after the pilot returned to the helicopter with the hiker, the helicopter was airborne again. According to the spotter, the pilot seemed to indicate that he intended to depart through a narrow path, a “tunnel in the clouds.” An interim risk assessment performed at this point may have indicated to the pilot that a different course of action would be more prudent. Even rudimentary consideration of the adverse weather conditions should have indicated to the pilot that it was no longer safe to take off and attempt to return to SAF at that time. At that point, the only safe option was to wait inside the helicopter at the remote landing site, contact SAR personnel for information and assistance, and wait for the weather conditions to improve. Although the temperature was near freezing, the helicopter provided goo minutes flying time from SAF, the pilot was likely very tempted to attempt to fly back to SAF rather than wait inside the helicopter for an indefinite period of time. The fact that the helicopter was airborne within about 9 minutes of the pilot’s return indicates that the pilot was still fixated on departing as soon as possible, and he did not spend much time considering alternative courses of action.106 Taking off in a helicopter in dark (moonless) lighting conditions, with marginal visibility, strong wind, turbulence, low clouds, the potential for structural and/or engine icing conditions, and surrounded by high terrain poses an unacceptably high risk of spatial and/or geographic 105 appropri G.A. Klein, eds., Linking Expertise and Naturalistic Decision Making (Mahwah, New Jersey: Lawrence Erlbaum Asso Although the dispatcher suggested that the spotter retrieve the lost hiker, the pilot, who was slightly more ately clothed for the conditions, hiked to retrieve her. 106 In this regard, the pilot’s decision-making performance during the accident mission is reminiscent of a category of decision-making error that researchers have labeled “plan continuation error.” Plan continuation error has been defined as “failure to revise a flight plan despite emerging evidence that suggests it is no longer safe.” [J. Orasanu, L. Martin, and J. Davison, “Cognitive and Contextual Factors in Aviation Accidents,” in E. Salas and ciates, 2001), pp. 209-225.] NTSB Aircraft Accident Report 50 disorientation, which could lead to loss of control and/or a controlled flight into terrain accident.

[ANALYSIS] The SAR mission extended into inet secretary, when he was a pilot in the NMSP aviation r was carried on board, and he routinely carried a survival g the area forecast for a local flight. TAFs and METARs are only valid within a 5-mile radius of the airport, but they provide detailed local weather information. It is possible, therefore, that the accident pilot reviewed the TAF and/or METAR for SAF before the accident mission. The TAF in effect for SAF at that time indicated that, between 1800 and 2200, temporary conditions would exist for the next 4 hours that included a cumulonimbus cloud ceiling at 6,000 feet agl (about 12,000 feet msl). (If such conditions extended an additional 15 miles to the northeast, there was a potential for thunderstorms and mountain obscuration in the search area during the mission.) The area forecast in effect for the high d light rain showers and isolated thunderstorms and gusting wind until 2100. A little more than 2 hours of daylight remained when the accident pilot was notified about the mission, and he might have thought it would be a short mission. The pilot knew that the hiker was in communication with the NMSP dispatcher via cellular telephone, and he might have believed that the hiker could quickly guide him to her location. This belief and the fact that the weather at SAF was warm and clear and the wind was calmer than earlier in the day likely contributed to the pilot’s decision to accept the flight. Further, the mountains in the search area were visible from SAF, and the pilot would have seen that they were clear of clouds at the time of departure. T d that the pilot predicted that the mission would be a quick “in and out” flight.) However, the hiker was unable to provide the pilot with much of the useful guidance that lost hikers can typically provide (such as describing her position relative to the sun, nearby landmarks, or terrain features), likely in part because of her limited proficiency w nighttime. According to the DPS cab section, a supply of cold-weather gea kit and extra blankets on such missions. Additionally, based on the elevation of the targeted search area (estimated to be 11,700 feet), the pilot should have anticipated that the helicopter would be operating near the 102 Las Vegas, New Mexico, is located about 40 miles east of SAF and 34 miles southeast of the accident loca . viewed before accepting the acci tion 103 The NTSB was unable to determine which specific weather reports the pilot re dent mission.

[ANALYSIS] In addition, a review of the reported weather conditions at DLG during some of the pilot’s previous passenger-carrying flights from the lodge to the fishing camp revealed that the ceiling and visibility conditions reported at the time of the accident were better than what was reported at DLG during the flight performed the day before the accident. The NTSB concludes that the weather conditions forecasted for and observed in the area on the day of the accident did not appear to be exceptional compared to the conditions that the pilot experienced on previous flights. Thus, the accident pilot’s decision to depart on the accident flight appears to have been based on his own assessment that the weather had adequately improved to the extent that he felt comfortable making the flight. 2.3.2 Situational Awareness and Spatial Disorientation Because of the likelihood of localized variations, the weather conditions encountered during the accident flight are not known. Also, the passengers’ descriptions of the weather vary. Three passengers stated that the flight was conducted below the cloud ceiling in VMC; however, two of those passengers fell asleep before the time of impact. Another passenger who first stated that he had no indication of weather subsequently changed his statement to indicate that he could see only “white-out” conditions outside the airplane. Based on interviews with other pilots who had flown with the accident pilot, continuing VFR flight into instrument meteorological conditions (IMC), either deliberately or inadvertently, would be considered uncharacteristic of the accident pilot. Had the visibility and ceilings for the accident flight been 5 miles or greater (similar to that which a Cessna 208 pilot reported in the area about 30 minutes before the accident), the rising terrain at Muklung Hills would have been visible to the pilot from the airplane’s last recorded Sky Connect position. However, had the visibility been 3 miles (similar to the 1455 DLG observation of 3 miles visibility with light rain), Muklung Hills and the peaks that define the wide and narrow passes to the west would not have been visible to the pilot from the airplane’s last recorded position. Also, if a 1,000-foot overcast ceiling existed (similar to the 1455 DLG observation), the peaks of Muklung Hills would have been obscured but the floor of the tundra may have been visible. In that case, the Muklung River, which crosses the tundra and flows through the wide pass, likely would have served as a prominent landmark. Although one passenger had noted that the pilot’s visual navigational methods appeared to include following streams, it is not known whether the pilot typically used the Muklung River as a visual reference to navigate to the wide pass. 49 NTSB Aircraft Accident Report 50 The accident airplane was equipped with a variety of avionics designed to assist the pilot with navigation, situational awareness, and terrain avoidance.

[ANALYSIS] He indicated that the GCI chief pilot stated that he did not instruct pilots on how to clean the windshields and that most pilots at the lodge used soap and water. 45 NTSB Aircraft Accident Report 2. Analysis 2.1 General The investigation determined that the pilot was certificated and qualified in accordance with Federal regulations. The accident airplane was operated under VFR and was not certified for flight under IFR (aircraft used in IFR operations are subject to altimeter system and altitude reporting equipment tests and inspections, and there was no record that the airplane received such inspections). The accident airplane, although not certified for flight under IFR, was well equipped with navigation and communications equipment. Examinations of the recovered engine, propeller, and airframe components revealed no evidence of any preimpact failures. A review of weather observations revealed that meteorological conditions that met the criteria for MVFR and intermittent IFR conditions were observed at DLG throughout the period of time surrounding the accident flight. Weather camera images from DLG and KNW indicated that conditions in the region varied but appeared to consist most often of overcast cloud bases of unknown height and occasional light rain; some lower clouds (beneath the overcast bases) and the occasional hint of sunshine could also be seen. An image taken by one passenger about 1356 showed good visibility and relatively high ceilings at the lodge, and a weather camera image from Lake Aleknagik (the camera location nearest the GCI lodge) showed an overcast cloud base but good visibility at that location about the time of the airplane’s departure. A Cessna 206 pilot who was flying 8 nm away from the accident site about 10 minutes before the accident indicated that the weather conditions in the area included ceilings about 600 feet msl and more than 5 miles of visibility under the overcast layer, or slightly better. However, another pilot who flew in the area near Muklung Hills earlier in the day stated that the weather was “moving all the time” and that visibility in the passes could change from good to poor from one hour to the next. Because of the likelihood of localized weather variations, the cloud conditions and visibility at the accident site at the accident time could not be determined. 2.2 Accident Sequence In the absence of any ATC communications; air traffic radar data; or data from any CVR, FDR, or other crash-resistant flight recorder for the accident flight, the sequence of events was determined primarily by analyzing the sparse position reports from the Sky Connect system, the limited data extracted from the nonvolatile memory of the digital engine instruments, the available weather information (which was limited because of the potential for localized variability), the information provided by the surviving passengers (only two of whom were awake at the time of the accident), ground impact evidence, and airplane crush damage. The Sky Connect system provided five position reports for the airplane at 3-minute intervals during the accident flight.

[ANALYSIS] The Safety Board believes that the captain's judgment J was faulty when, in violation of FAR Part 91, he chose to continue WFP flight into ! bs i IMC during climbout in an area of high terrain. }} 4 I 2.3 The Captain's Experience and Judgment The captain was certificated and qualified in accordance with applicable FARs. The captain's log book was not available for verification of his aeronautical experience; however, a reconstruction of his flying experience from previous employers and FAA records indicated that he did not possess the experience required by the company operations manual. It is common practice in the industry to use flight hours as an indicator ' of aeronautical skill. Pilot flight hours are a universal measure of pilot experience . and competence, and they play a role in evaluating a pilot's ability to make sound , aeronautical decisions. The investigation disclosed that the captain had significantly misrepresented his professional credentials conceming his flight experience, training, and employment on resumes and employment applications. As a result, several employers dismissed or rejected the captain when his aeronautical skills failed to meet qualifications and/or performance standards for various pilot positions. 37 The Safety Board believes that the judgment of the captain to continue VFR flight into IMC rather than to practice appropriate weather avoidance techniques resulted in a collision with obscured mountainous terrain. This decision demonstrates a lack of appropriate aeronautical judgment skills and is a reflection of insufficient professional training and experience. SAT used an employment application and a resume, which contained false information, to evaluate the captain's professional background and experience and did not attempt to verify the information provided. At the time the captain was employed, he did not meet SAT's criteria of 2,500 total hours and 1,000 multiengine hours of flight experience for a pilot position. Furthermore, the captain had not met these requirements at the time of the accident. The circumstances of this accident and the Safety Board's previous accident investigation experience have demonstrated the consequences of poor judgment and poor decision making by pilots. The FAA and other aviation industry organizations have supported projects that have resulted in the development of Aeronautical Decision Making (ADM) training materials aimed at improving a pilot's ability to recognize and control hazardous thought processes and situations.

[ANALYSIS] The wreckage condition indicated a trajectory with little forward motion and high vertical impact forces. The wreckage pattern is consistent with ground contact in a stalled condition. The Safety Board believes that maneuvering into such a flight attitude would not have been necessary or attempted if the flight had been conducted in visual meteorological conditions as required for VFR flight. VFR requires a minimum flight visibility of 3 statute miles and various distances” from clouds. If these conditions had existed along the flightpath, the mountainous terrain Jeading to the crater would have been visible and avoided by the pilot during his climb toward Maui. 2.2 The Flight The Safety Board tried to determine why the pilot, after passing Upolu Point, deviated from both his intended flight plan and his stated intention to overfly R-3104, Why he did not circumnavigate clouds that presented less than VMC over Maui is also unclear. The track that he was observed to fly, 300- to 321-degrees magnetic, lead directly toward the high terrain, which is one of the most prominent landmarks in the Hawaiian Islands. SAT pilots were well aware that they were authorized to conduct operations only in VMC and to deviate from designated routes only to the extent necessary to avoid weather. Consideration was given to the possibility that the pilot intended io retum to Kahului Airport (Maui) to board the SAT employee who was dropped off during the earlier morning flight. However, the captain's VFR flight plan direct to HNL and the clearance to overfly the Island of Kahoolawe (R-3104) at 6,500 feet indicate that he planned to proceed on a more westerly and direct course. The restricted area is south of Maui and the company-designated return route. The captain's intention to navigate well south of Maui seems reasonable considering the weather that was affecting the southeastern portion of the island. The Safety Board concluded that a retum flight to Kahului Airport was not intended. Another possibility considered was that the pilot diverted from the standard route of flight to show the Mount Haleakala volcano crater to his 9Basic VFR weather minimums are contained in FAR 91.155. The minimum distances from clouds for the accident flight were 500 feet below, 1,000 feet above and 2,000 feet horizontal. 33 passengers. Postaccident observations indicate that the weather along the southeastern slope and the summit of the crater was not suitable for visual sightseeing activity. The captain would have been aware of this fact upon departing Upolu Point as he surveyed the horizon to the west and northwest. Consequently, this possibility was discounted because of the weather in the Haleakala area. Investigators also considered the possibility that passengers attempted to lure pilots with gratuities to deviate from their intended route; however, such a practice could not be substantiated. Also at the time of his departure from Hilo the pilot was aware that upon deplaning his passengers in HNL, he was to fly to Maui to board additional tourists and return them to HNL. Consequently, the captain knew that adhering to his intended route schedule was a necessity.

[PROBABLE CAUSE] Vogt Chainnan Vice Chairman John Hammerschinidt Member Christopher A, Hart Member February 2, 1993 John K. Lauber, Member, filed the following dissenting statement: I have long been on record that I believe our probable cause findings are primarily a vehicle for effecting positive changes, and not for placing blame. In accident investigation and prevention efforts, I don’t believe that we are constrained to a narrow construct of causality. By embracing a “pilot error" probable cause, as it has in this case, the majority has, in my opinion, foregone an important opportunity to leverage meaningful changes that would be more helpful in the prevention of future accidents like this one. The safety message in the probable cause as adopted by the majority is minimal to nil: Pilots should not make errors, especially grievous errors such as continuing VFR flight into instrament meteorological conditions. Because this pilot's performance was so egregious, I venture to say that few pilots will see any apparent relationship between what we believe this pilot did and his or her own piloting skills. Such denial is an especially potent force among those pilots whose character and judgment flaws would lead them to take risks similar to what this pilot did; those who need to hear this message the most are the least likely to gain any meaningful insight into their own behavior from the probable cause adopted by the majority. It is a fact that among the population of pilots, there are some who do not possess those qualities of character and judgment so necessary to be a safe pilot. Even though they may possess the technical qualifications, ie., the proper Sl certificates, these are not the people to whom: the flying public should have to entrust their lives. Many times these flaws are very difficult to identify in a timely fashion. In this case, however, this pilot left a readily identifiable trail of information that indicated he was not likely to exercise the degree of care and caution we demand of professional pilots. Our investigation revealed that this pilot had been dismissed by five employers for misrepresentation of qualifications and experience, failure to report for duty, disciplinary action, poor training performance, and work performance that was below standards. Based on a background check, another operator rejected his application for a pilot position for failing to disclose information and misrepresentation conceming previous employment. Scenic Air Tours apparently conducted no such extensive background check, and as a result eight paying passengers were entrusted to this pilot's care. It is certainly true, as the majority holds, that this pilot's actions were directly causal to this accident.

[ANALYSIS] Although the pilot was not required to use these navigational aids, when he departed visual meteorological conditions and flew into IMC, he should have used the navigational aids to accurately monitor his ground track and altitude. During the last few minutes of flight, when the airplane’s ground clearance 12 NTSB/AAB-01-02 was rapidly decreasing, the pilot did not reverse course or take emergency action. Radar data indicates that at this point in flight, the airplane’s track varied little from its predominately westnorthwesterly direction. If the pilot had been using his navigational aids correctly, he would have realized that he was nearing high terrain and would likely have changed his course. The pilot’s early awakening time and the length of his duty day raise the possibility of fatigue as a factor in this accident. However, in the absence of further evidence, the Safety Board cannot conclusively determine whether fatigue was a factor in this accident. Because no blood sample from the pilot was available for analysis, the impact of the drug phentermine on pilot’s behavior could not be analyzed or determined. PROBABLE CAUSE The National Transportation Safety Board determines that the probable cause of this accident is the pilot’s decision to continue visual flight into instrument meteorological conditions (IMC) in an area of cloud-covered mountainous terrain. Contributing to the accident were the pilot’s failure to properly navigate and his disregard for standard operating procedures, including flying into IMC while on a visual flight rules flight plan and failure to obtain a current preflight weather briefing. Adopted on September 26, 2001

[ANALYSIS] ANALYSIS The pilot was properly certificated and qualified in accordance with applicable Federal regulations and company requirements. The airplane was properly certificated and equipped in accordance with Federal regulations and approved procedures. No maintenance discrepancies or corrective actions were noted between the airplane’s last scheduled maintenance activity on August 31, 1999, and the date of the accident. The airplane had no recently reported maintenance history of navigation-related problems. Surveillance of Big Island Air by the principal operations inspector (POI) was consistent with the Federal Aviation Administration’s (FAA) existing guidelines. On August 30, 1999, 25 days before the accident flight, the pilot completed a recertification training program after his 11-month leave of absence from Big Island Air’s employment as a tour pilot. During the FAA-administered flight check, the POI reviewed with the pilot the requirements of Special Federal Aviation Regulations 71, which provides operating rules for visual flight rules (VFR)-only air tour flights conducted in Hawaii under 14 Code of Federal Regulations Part 135. Remarks on the “Airman Competency/Proficiency Check FAR [Federal Aviation Regulations] 135” form indicate that the pilot completed the requirements satisfactorily. Big Island Air’s FAA-approved Operations Specifications and corresponding training program clearly indicate that all tour flights were to be conducted under VFR; no flying under instrument flight rules was authorized at any time. The accident pilot had over 11,500 hours of flight time, most of which was accumulated in the Hawaiian Islands. The pilot was therefore likely aware of and understood the VFR flight visibility and cloud clearance limitations. However, on September 4, 1999, 5 days after his recertification training, the accident pilot performed a tour flight, which was videotaped by a passenger. During the flight, the airplane can be seen flying in clouds on several occasions and in different locations. Further, witnesses indicated that on the day of the accident, the sky was overcast in the vicinity of the accident site. The Safety Board therefore concludes that on the accident flight, the pilot flew into instrument meteorological conditions (IMC). The pilot was responsible for but did not obtain a preflight weather briefing from the FAA’s Honolulu Automated Flight Service Station as required by the FAA. The FAA has no record of the pilot requesting a weather briefing for the accident flight or the flight he conducted earlier that day. The Safety Board concludes that the pilot’s failure to obtain a preflight weather briefing was a deviation from standard operating procedures. The accident airplane was equipped with very high frequency omni directional range, distance measuring equipment, and global positioning satellite receivers, which could have been used to indicate the accident airplane’s position and ground clearance. Although the pilot was not required to use these navigational aids, when he departed visual meteorological conditions and flew into IMC, he should have used the navigational aids to accurately monitor his ground track and altitude. During the last few minutes of flight, when the airplane’s ground clearance 12 NTSB/AAB-01-02 was rapidly decreasing, the pilot did not reverse course or take emergency action.

[ANALYSIS] Thus, the NTSB concludes that the PIC’s decision to continue the VFR flight into reduced visibility conditions resulted in the flight entering IMC. The NTSB further concludes that the investigation found no evidence that NTSB Aircraft Accident Report 44 management or scheduling pressures, habitual noncompliance with company policy, or history of risk-taking behaviors influenced the PIC’s decision to continue the flight. Although the SIC was a relatively low-time pilot and new to flying in the area, the need to avoid and escape an inadvertent encounter with IMC when flying under VFR is reinforced in all levels of pilot training. Considering the SIC’s reported perception that some Hageland pilots operated in reduced visibility, it is possible that the SIC believed such operations were acceptable. Although the PIC had ultimate responsibility for the flight and should have recognized that weather conditions were deteriorating and taken action to avoid encountering IMC, the SIC also had a responsibility to speak up and challenge the PIC if he observed unsafe conditions. 2.3.2 Failure to Escape IMC Hageland’s CFIT-avoidance procedures outlined in its CFIT-Avoidance Training Manual stated that, if IMC is encountered, the pilot must take immediate action, using his or her best judgment in executing any maneuver required to exit IMC. Per the manual, these maneuvers could include using instrument references to reverse course to return to VMC or executing a high-performance climb to a safe altitude and requesting an IFR clearance. The airplane was equipped with a Garmin GMX 200 MFD that provided color-coded terrain information and advisory flags to assist the flight crew with terrain awareness; however, the MFD operated independently from the TAWS and, per its manual, provided only advisory information that was not to be used for navigation. According to the manual, pilots flying TAWS-equipped airplanes are expected to execute emergency actions in response to warnings (see sections 2.3.3 and 2.4.2 for further discussion about the flight crew’s TAWS use and CFIT-avoidance training, respectively). In a two-person crew, the PIC’s duty and authority included setting the tone in the cockpit and demonstrating good leadership and adherence to standard operating procedures (SOPs), regardless of which pilot was the PF. The PIC’s CRM skills were particularly important for this crew pairing because the SIC was new to Hageland and would be looking to the PIC to set the expectations for the flight and, in a broader sense, the culture at Hageland. According to Hageland’s chief pilot, CRM was the “cornerstone of good judgment and decision-making.” Thus, the NTSB concludes that, once the flight entered IMC, the PIC should have either executed an escape maneuver or commanded the SIC to execute one. The NTSB has long recognized the importance of effective CRM in accident prevention.

[ANALYSIS] The nearest weather observing equipment (at PATG, about 10 nm southeast of the accident site at an elevation of 18 ft msl) reported conditions that met the criteria for VFR about the time of the accident. However, weather for the area can change over very short distances in variable terrain; thus, the observations at PATG, particularly the ceiling and visibility measurements, likely were not applicable to the accident site. Several sources of weather information indicated that the regional environment was suitable for decreased visibility and clouds below 2,000 ft msl at the time of the accident. These included reports from PATG of light rain and a small temperature/dew point spread and images from FAA weather cameras at PATG that showed, at times, terrain 7 miles west of the camera was completely obscured and terrain 6 miles west was partially obscured. In addition, weather radar imagery identified hydrometeors in the area above the accident site at the accident time, which could have been reaching the surface in the form of light rain. A second company flight crew, who also departed PAQH for PATG in a Cessna 208B about 2 minutes after the accident flight crew, chose to divert around the mountainous terrain near the location of the accident site to avoid clouds. Data for this second flight showed that, before it diverted at 1156, its route and altitude were similar to those of the accident flight. Based on the accident flight’s last data point (location and time) and the location and estimated time of the accident, the accident flight crew had likely passed, about 5 minutes earlier, the location where the second crew chose to divert. The second flight crew, after landing in PATG and learning of the accident airplane’s ELT signal, departed at 1231 to try to locate the accident site. They arrived at the area of the ELT signal within an hour of the accident but were unable to see the accident site due to clouds obscuring the mountain. Thus, the NTSB concludes that, based on the available weather information that indicated the likelihood of decreased visibility due to precipitation and/or clouds in the accident NTSB Aircraft Accident Report 43 area and the observation from a company flight crew that clouds obscured the accident site within an hour after the accident, the accident flight crew likely encountered IMC before the collision with terrain. 2.3 Flight Crew Performance Although most of Hageland’s Cessna 208 flights operated under VFR were single-pilot operations, the company sometimes assigned an SIC to assist with cargo or to extend the allowable flight time per duty period. The company could also assign a safety pilot to act in a supervisory role to ensure a new PIC was following procedures. The DO stated that an SIC was expected to act as a second crewmember. As an active crewmember, it is possible that the accident SIC was assigned the role of PF for the accident leg.

[ANALYSIS] The RA applied to all five proposed flight segments and considered a variety of risk factors, including weather conditions along the flight route, aircraft equipment, and whether the proposed flight would fly under IFR or VFR. When assessing the flight, the OCA initially recommended that the PIC depart under IFR due to rain he observed near PAQH. However, after further discussion, the OCA agreed with the PIC’s assertion that, based on the available weather information, the flight could depart under VFR in compliance with company policy and regulations. The OCA subsequently released flight 3153 as a VFR flight with an agreed-upon risk level of RA 2, which was a cautionary risk category that did not require management review. Based on the form’s criteria, if the accident flight had been designated an IFR flight, the risk category would have remained RA 2 and, therefore, would not have required management review. NTSB Aircraft Accident Report 42 2.2 Accident Sequence The accident occurred during the flight crew’s third flight segment of the day, which departed from PAQH at 1133 en route to PATG. According to the airplane’s Spidertracks data, after departure, the flight proceeded along a generally direct route toward PATG. A TAWS simulation for an estimated accident flight that assumed that the airplane maintained cruise flight about 1,000 ft msl until the last known data point showed that the corresponding terrain clearances would have been between 500 and 700 ft agl. The TAWS simulation showed that the accident airplane’s TAWS, if not inhibited, would have provided nearly constant alerts (see section 2.3.3 for additional discussion). Hageland had no designated day VFR flight route between PAQH and PATG. The GOM specified that flights be planned along the shortest safe route or as assigned by ATC and that VFR flights be flown at altitudes no lower than 500 ft agl. According to the Hageland chief pilot and a safety pilot who both testified at the investigative hearing, the company encouraged pilots to fly at higher altitudes; however, pilots often operated under a ceiling when visibility was “really good” below the ceiling, and flights below 1,000 ft agl were not uncommon. In the final 4 minutes of the flight (after the last recorded data point at 1,043 ft msl), the airplane’s altitude increased to about 2,300 ft msl, which was the elevation of its initial impact with the mountain ridge (section 2.3.3 describes the TAWS simulation for this part of the flight). Wreckage damage was consistent with an extreme nose-up attitude at the time of the collision. The nearest weather observing equipment (at PATG, about 10 nm southeast of the accident site at an elevation of 18 ft msl) reported conditions that met the criteria for VFR about the time of the accident.

[ANALYSIS] All of these risk factors were present for the pilot during the last minute of the accident flight. Attempts to continue visual flight into IMC may be even more problematic for helicopter pilots than for pilots of fixed-wing aircraft because helicopters are inherently less stable and require near continuous control inputs from the pilot. Helicopters, like the accident helicopter, that are not equipped for IFR flight and do not have control stabilization or autopilot systems can impose high perceptual and motor demands on the pilot. These helicopters can make it very challenging for pilots to maintain stable flight by referring to flight instruments alone. When the accident pilot attempted to continue visual flight into IMC, he would have been subjected to a high workload in order to maintain control of the helicopter. Moreover, the extent of the weather echoes and the duration of the sustained turn captured by the radar track suggest that the encounter with IMC was Analysis 38 Aircraft Accident Report prolonged. This would have further complicated the pilot’s workload and increased the potential for spatial disorientation resulting from hazardous illusions, thereby, increasing the potential for inappropriate control input responses. In the absence of evidence indicating a mechanical malfunction, severe turbulence,89 or some other factor that would explain the accident pilot’s apparent loss of control of the helicopter, spatial disorientation is the most likely explanation. The Safety Board concludes that the helicopter’s descending spiral flightpath, which became increasingly erratic in the final seconds of the flight, was consistent with pilot spatial disorientation. 2.3.2 Pilot’s Decision-making Although the pilot likely encountered areas of deteriorating weather and IMC near the accident site, photographs recovered from a passenger’s camera showed areas of clearer weather along the coastline, east of the helicopter’s position, as it exited Waialeale Crater. The lower terrain between the helicopter and the coastline and the height of the clouds would have permitted a safe deviation in that direction. Furthermore, Bali Hai pilots were authorized to fly as low as 1,000 feet agl from the crater toward the coast; thus, the pilot could have flown in the direction of better weather, while remaining in compliance with SFAR 71 and the certificate of waiver or authorization requirements. Rather than deviate to the east, however, the pilot chose to fly south, along his usual tour route adjacent to Kahili Ridge, as the advancing line of clouds closed in on the high terrain. The pilot’s decision to fly into an area of deteriorating weather and high terrain, despite the availability of a safer route to the east, was an error that played an important causal role in this accident, and the Safety Board considered many factors that may have influenced that decision. 2.3.2.1 Inexperience with Local Weather Patterns The accident pilot had been flying commercial air tours for Bali Hai on Kauai for less than 2 months. Nearly all of his previous flight experience had been gained as a military pilot with the Indian Air Force.

[ANALYSIS] Nearly all of his previous flight experience had been gained as a military pilot with the Indian Air Force. Though the pilot was experienced in helicopter operations and claimed previous mountain and coastal flying experience, he had few flight hours as a commercial air tour pilot and had limited knowledge of Kauai’s weather patterns. The pilot had no previous experience flying in Hawaii before he was hired by Bali Hai, and he began conducting tour flights after accruing just 6.7 hours of flight training from company personnel, none of which included specific training on Kauai weather. Many experienced local air tour pilots interviewed after the accident stated that VFR helicopter operations on Kauai were unusually challenging because of the rugged terrain, mountain winds, and rapidly changing visibility and cloud conditions. They stated 89 The forecast included only moderate turbulence, and the weather study found no evidence to indicate a likelihood of severe turbulence. Analysis 39 Aircraft Accident Report that these conditions rendered traditional sources of pilot weather information, such as automated reporting stations and FSS briefings, not very useful, and that this heightened the importance of a pilot’s skill in visually assessing changes in weather conditions during tour flights. The experienced pilots indicated that a pilot’s skill in assessing changing weather conditions and anticipating the effect of any changes on flying conditions was critical for effective decision-making. They stated that such skills improved as local flying experience increased. This raises concern about the impact of local inexperience on the safety of pilot decision-making. A review of the eight weather-related air tour accidents that have occurred in Hawaii since SFAR 71’s implementation revealed that four of the accidents involved pilots who had relatively low experience flying air tours in Hawaii. In fact, three of the accidents involved pilots who had flown there for less than 2 months. It is highly unlikely that the accident pilot would have decided to continue into the area of deteriorating weather conditions and attempted to cross in the vicinity of the accident site if he had accurately assessed the changing weather and had appreciated how it would likely affect flight visibility in those areas. Therefore, the pilot’s decision to fly in the vicinity of the accident site indicates that he was unable to accurately assess how rapidly or to what extent the weather was deteriorating in that area. As described earlier in this report, the pilot had previously told a passenger that he had flown through thin clouds during a ridge crossing because he could see through the clouds. The pilot may have expected that this would be the case during the accident flight; however, visibility was dramatically reduced along the top of the ridge as the incoming line of weather reached the high terrain. The Safety Board concludes that the pilot’s inexperience with Hawaii weather conditions affected his ability to make appropriate in-flight decisions when faced with deteriorating weather.

[ANALYSIS] Although these lapses of controller performance are cited as causal, the Safety Board also has chosen to recognize that the controllers’ performance was a direct product of FAA air traffic management institutional decisions and practices that do not allow for human performance lapses in judgement or decision making. The air traffic control procedures permitted the local controller to assume full and complete responsidility for in-trail separation of aircraft on the final approach by invoking visual separation standards. As a result, two critical problems arose: (1) the separation distance between EA 111 and N44UE was reduced from the radar requirement of 2.5 miles minimum {{inside the final approach fix) to some indeterminate 35 distance necessary for N44UE to clear the runway prior to the arrival of EA 111 over the threshold; (2) an important redundant element, (the monitor controller) was removed from the loop. In addition, the air traffic procedures allow for the issuance of multiple landing clearances, which were issued in this case to CO 9687, N44UE and EA 111 in a period of 49 seconds while all three aircraft were still on final approach. An effect of this action was to remove another redundant element in the system: a1] aircraft had their landing clearances, and therefore no further communications with the local controller were necessary. When the simple but compelling distractions caused the local controller to divert his attention away from the landing aircraft, the stage was set for this accident. It is well-documented that human performance is subject to simple lapses (errors of omission), particularly in the presence of distracting events. Thus, the designers and operators of complex systems, who implement design features and operating procedures that allow one individual to assume the full burden for safety-critical operations, like the Air Traffic Control system, must bear some of tne responsibility for those accidents attributed to the occasional lapse in the performance of a single individual. Therefore, in addition to noting individual performance in the assessment of causal and contributing factors, the Safety Soard cites the broader failure of the Federal Aviation Administration to provide ATC equipment and prcecedures that adequately take into consideration those occasional tapses in human performance that must be expected. In examining the specific circumstances of this accident, the Safety Board became concerned that the current provisions in ATC procedures permit the controller to issue Yanding clearances to several aircraft in succession without assurance that the adequate separation will be maintained among those aircraft as they approach the landing threshold. Correlation of ATC communications and radar position data indicates that CO 9687’s clearance to land was given when the airplane was about 1 mile from the threshold of runway 26 right and the landing clearances to N44UE and EA 111 were given when the former was about 1/2 mile outside the final approach fix (FAF) and when the Jatter was about 3/4 mile outside of the FAF, distances of about 4.5 miles and about 6 miles, respectively, from the threshold.

[CONCLUSIONS] The Eastern flightcrew had three opportunities to learn about preceding landing traffic by listening to the tower frequencies; however, the time between transmissions, the large number of transmissions, and the required duries in the cockpit would have limited the utility of that information. Airspeed reduction transmissions to N44UE by the radar monitor controller were insufficient to achieve the required 4 miles separation from the preceding airplane, CO 9687, on the final approach and at the threshold. The absence of appropriate airspeed reduction instructions to EA 111 by the Atlanta approach north final and the radar monitor controllers led to a speed differential that resulted in a loss of the separation between EA 11] and N44Ué. The traffic volume at the time of the accident presented an average controller workload, but the local controller was distracted with radio difficulties (misunderstood instructions) when communicating with CO 9687. 42 The local controller’s distraction by communication difficulties with CO 9687 was prompted by his perceived need to clear runway 26 left for another airplane inbound with a hydraulic emergency and the possibility of a runway incursion from CJ 9687 during taxiing. The controller was inattentive to the more immediate task of monitoring the separation of traffic landing on runway 26 right. 3.2 Probable Cause The National Transportation Safety Board determines that the probable causes of this accident were (1) the failure of the Federal Aviation Administration to provide air traffic control procedures that adequately take into consideration human performance factors such as those which resulted in the failure of the north local controller to detect the developing conflict between N44UE and EA 111, and (2) the failure of the north local controller to ensure the separation of arriving aircraft which were using the same runway. Contributing to the accident was the failure of the north local controller to follow the prescribed procedure of issuing appropriate traffic information to EA 111, and failure of the north final controller and the radar monitor controller to issue timely speed reductions to maintain adequate separation between aircraft on final approach. 4. RECOMMENDATIONS Therefore, as a result of the investigation of this accident, the National Transportation Safety Board recommends that the Federal Aviation Administration: Develop an Air Traffic Bulletin and provide a mandatory formal briefing to all air traffic controllers on the importance of, and the need for, giving traffic information when issuing an anticipated separation landing clearance. (Class II, Priority Action) (A-91-27) Amend the Air Traffic Control Handbook, 7110.65F, paragraph 3-127, to preclude the issuance of multiple landing clearances to aircraft outside of the final approach fix.

[ANALYSIS] No indication was given that they were number two for landing behind another airplane. The Safety Board believes that flightcrews are conditioned to receive such information. as required in the Air Traffic Con*rol Handbook procedures relating to anticipating separation. If the controller had provided traffic information to the EA 11] flightcrew, the flightcrew’s sense of situational awareness and motivation to search for a preceding airplane might have been increased. Lacking such information, it appears that the crew proceeded through their normal task of completing a routine night landing on a runway to which they had been cleared, unaware that there was another airplane on the runway. The fact that EA 111 had received a landing clearance did not relieve the flightcrew of responsibility to "see and avoid" other aircraft in their vicinity. However. in the absence of conspicuous lighting on the King Air and without prompting from ATC to direct their attention to traffic ahead, it was extremely difficult, if not impossible for the EA 11] flightcrew to detect the other aircraft on the runway. Moreover, there is a concept known as diffusion of responsibility that describes a tendency on the part of pilots in some circumstances to relax their vigilance. A National Aeronautics and Space Administration study on near midair collisions‘ sBillings, C., Greyson, ®., Hecht, W., and Curry, R., "A Study of Near Midair Collisions in U.S. Terminal Airspace," NASA Technicat Memorandum 81225, 1980. 32 indicates that an inappropriate sense of shared responsibility may occur when an airplane is under ATC radar control. In such a circumstance, a pilot may relegate a portion of his responsibility for vigilance to the controller for seeing and avoiding other aircraft. In the case of EA 111, having come from the radar environment of the approach and after having received specific landing clearance, the pilots may have experienced a natural tendency to relax in their attempts to visually search for an aircraft between their position and the intended landing runway. In any event, the Safety Board found no evidence to indicate less than expected vigilance by the EA 1}}] flightcrew. The Safety Board thus concludes that the actions of the EA 111 flightcrew, while not optimal in terms of speed control and situational awareness during initial and final approaches, were not uncommon to airline operations and were not causal to the accident. 2.4 Role of Air Traffic Control The Safety Board also evaluated the performance of the air traffic control personnel involved in this accident. The final controller was responsible for maintaining separation of succeeding airplanes on the approach to the outer marker. The monitor controller was responsible for maintaining separation of succeeding airplanes on the approach from the outer marker to within 1 mile of the runway.

[ANALYSIS] At the time of the north local controller’s transmission, EA 111 was almost 6 miles from the runway and the King Air was about 3 miles out. However, the distance between the two aircraft was decreasing at an unacceptable rate and was less than the required 2.5 miles separation as N44UE arrived at the runway threshold. The north local controller, in an attempt to maintain the lancing sequence, initiated visual separation between N44UE and EA 11]. At the time visual separation was initiated, the required minimum radar separation standard of 2 1/2 miles did exist between N44UE and €A 11:. However, to make Sure that an approved separation standard would exist after using visual separation, the local controller would have had to monitor both airplanes closely to assure that EA 11] did not cross the runway threshold until N44ue had been observed leaving the runway (Air Traffic Control Handbock 7110.65, paragraph 3-122, Same Runway Separation}}. Unfortunately, EA 111 had about a 45 knot closure rate on N44UE, and ATC radio transcripts indicate that no action was taken to reduce the rate of closure. In addition, the radio transcripts indicate, and a personal interview confirmed, that the local controller became distracted by radio communication difficulties with the flightcrew of CO 9687. The local controller stated to investigators that he did not observe the touchdown and rollout of N44UE at the runway threshold or during landing. The Safety Board reviewed the pertiient FAA Air Traffic Control Handbook 7110.65F requirements and concluded that the instructions contained therein clearly define the controller’s responsibilities for “same runway separation" and “anticipating separation.” The Safety Board concluded that the physical evidence on the runway and on both airplanes indicated that the collision occurred on a runway that was the responsibility of the narth local controller. The Safety Board attributes the north local controller’s distraction and preoccupation with efforts to communicate with CO %687 to his gerceived need to clear runway 26 left for another airplane inbound with a hydraulic emergency. However, the Board believes that the local controller’s concern that the flightcrew of CO 9687 was going to cross runway 26 ieft without a clearance was not well founded. At 1903:03 radio transmissions indicate communications difficulties between CG 9687 and the tower regarding its taxi instructions. They transmitted, "tower, Continental ninety six eighty seven bravo two [taxiway] holding short." The fact that the airplan7 was on the taxiway and not moving was substantiated by the ground controller in the tower cab. The CO 9687 fligh:crew took positive action to avoid becoming a hazard. They stopped clear of the active runways and remained in position until their clearance was clarified. 34 The Safety Board conciuded that there was a lack of understanding between the tocal controller and CO 9687, and that this communications anomaly did not result from any oquipment failure.

[ANALYSIS] The final controller was responsible for maintaining separation of succeeding airplanes on the approach to the outer marker. The monitor controller was responsible for maintaining separation of succeeding airplanes on the approach from the outer marker to within 1 mile of the runway. It is evident, by the airspeed reductions that were issued by the monitor controller to the flightcrew of N44UE, that he was attempting to achieve additional separation between CO 9687 and N44UE prior to N44UE crossing the runway threshold of 26 right. The recorded radar data indicate that the ceparation between CO 9687 and N44UE never exceeded 3.5 miles. Therefore, the monitor controller’s action failed to achieve the 4 mile minimum required separation standard. He also failed to compensate for the added closure rate that occurred between N44UE and the following airplane, EA 111, as a result of the airspeed reductions he issued to N44UE. EA 11] was about 4.0 miles behind N44UE at FREAL intersection. In order to accomplish the desired sequencing of EA 111 trailing N44UE, an early speed reduction for EA 111 was required. A timely and sufficient airspeed reduction adjustment was not issued by either the final controller or the monitor controller. The required separation between EA 111 following N44UE was 2 1/2 miles inside the final approach fix. It appears that the monitor controller was late in recognizing the potential conflict of decreasing separation between N44UE and EA 11). About 6 miles from the runway, he assigned an airspeed change to EA 111, "reduce to your final approach speed." This speed assignment was not in conformance with the Air Traffic Control Handbook, which states that a controller shall advise an aircraft to increase or decrease to a specified speed in knots. In addition, the monitor controller did not receive an acknowledgement from the flightcrew of EA 11) for the instruction to reduce to approach speed, and thus should not have assumed that the instruction had been received and complied with. Therefore, 33 the monitor controller initiated a sequence of events that caused the final approach interval spacing to quickly approach the minimum of 2 1/2 miles. Although he was relieved of direct responsikility for the ensuing loss of separation when the north local controller transmitted "EA 111, you are in Sight, cleared to land 26 right," the Safety Board believes that the monitor controller’s action contributed to the speed differential and to the overtake that ultimately was a factor in the accident. At the time of the north local controller’s transmission, EA 111 was almost 6 miles from the runway and the King Air was about 3 miles out. However, the distance between the two aircraft was decreasing at an unacceptable rate and was less than the required 2.5 miles separation as N44UE arrived at the runway threshold.

[ANALYSIS] The CO 9687 fligh:crew took positive action to avoid becoming a hazard. They stopped clear of the active runways and remained in position until their clearance was clarified. 34 The Safety Board conciuded that there was a lack of understanding between the tocal controller and CO 9687, and that this communications anomaly did not result from any oquipment failure. Rather it was the result of an incomplete transfer of intormation (taxi instructions) between the controller and the flightcrew of CO 9867. The net effect of this lack of information transfer was to create a self-imposed workload on the controller that was sufficiently high to cause him to disregard other higher priority tasks. If the local controller had been so concerned that the airplane was going to cross runway 26 left without a clearance, he had the option of discontinuing departures from that runway and directing ASE 301, that was holding in the takeoff position, to clear runway 26 left. As long as ASE 301 was holding in position on the runway, it could not be threatened by a possible runway incursion from an airplane at the opposite end of the runway. However, the local controller became distracted for a critical period by the possibility of a runway incursion involving ASE 301 and CO 9687. Eventually, et 1904:13, the local controller cleared ASE 301 for takeoff on runway 26 left. The collision of €A !1] and N44UE had taken piace on runway 26 right at 1904:07. It is recoynized that concentration on one task can overload a person to the extent that other relevant cues are disregarded or otherwise not attended to, leading to a degradation of overall task performance. Because the north local controller focused his attention on the path of CO 9687 on taxiway Bravo, at the west end of runway 26 left, he was distracted at a critica) time from the landing rollout of N44UE and the FA 111 airplane that was about to cross the threshold and land on the same runway. The Safety Board concludes that this accident was a result of lapses in the performance of the Atlanta tower north tocal controller and, to a lesser extent, the performance of the Atlanta approach control north final controller and the radar monitor controller. Specifically, the north local controller did not ensure the separation of the aircraft approaching and landing on runway 26 right. further, ne failed to follow the prescribed procedure of issuing appropriate traffic information to the crew of EA 11). This information would have improved the flightcrew’s situational awareness and their motivation to search for the preceding King Air. The Atlanta approach north final controller and the radar monitor controller had opportunities to issue timely speed reductions to ensure that adequate separation was maintained between the successive aircraft on final approach, but did not do so. Although these lapses of controller performance are cited as causal, the Safety Board also has chosen to recognize that the controllers’ performance was a direct product of FAA air traffic management institutional decisions and practices that do not allow for human performance lapses in judgement or decision making.

[ANALYSIS] Provision of the multiple clearances to land in retatively rapid succession may have provided the north local controller with time needed to devote attention to flights waiting for takeoff clearances from runway 26 left. However, the premature clearances also make the north local controller the only controller responsible for the spacing between successive flights by removing that responsibility from the monitor controller and it may also have reduced the vigilance of the flightcrew. As a result, appropriate spacing for completion of the landings depended entirely on the continued vigilance of the local controller and the flightcrew. In this case, the flightcrew of EA 111 probably could not have seen N44UE because of N44UE’s external lighting configuration. Also, the premature clearance for EA 111 to land removed the redundancy of flightcrew vigilance when the north local controller subsequently became distracted with 36 CO 9687’s taxi clearance. Without the clearance to land, EA 111 would have had to remind the local controller that the clearance was needed. This reminder would probably have redirected the controller’s attention to tte lack of adequate spacing between N44UE and EA 11] and may have led to a correction of the problem by denying EA 11] clearance to Jand. In the 1977 edition of the Air Traffic Control Handbook, 7110.65/, the issuance of multiple landing clearances was not allowed. Specifically, paragraph 1122, “Anticipating Separation,“ stated, “Landing clearance need not be withheld until prescribed separation exists if there is reasonable assurance it will exist when the aircraft crosses the landing threshold. However, do not clear a succeeding aircraft to land on the same runway before a preceding arriving aircraft crosses the landing threshold...." This is basically the same text that is contained in the current Handbook, 7110.65F; however, the earlier procedures went on to say, "...do not clear more than the first two aircraft to land at any one time ai include traffic information with the clearance." During March 1978, isis paragraph was changed to delete the numerical limits for clearing aircraft to land. The Safety Board believes that current ATC procedures, as they oertair to the anticipated separation of arriving aircraft, require nearly flawless human performance that makes no allowance for an error of omission or lepse of attention due to any type of distractive event. Therefore, the Safety Board believes that the procedures contained in the Air Traffic Control Handbook, 7110.65F, paragraph 3-127, "Anticipating Separation," should be amended to preclude the issuance of multiple janding clearances to aircraft outside of the final approach fix.

[ANALYSIS] Therefore, the Safety Board believes that the procedures contained in the Air Traffic Control Handbook, 7110.65F, paragraph 3-127, "Anticipating Separation," should be amended to preclude the issuance of multiple janding clearances to aircraft outside of the final approach fix. Also, a numerical limit should be established so that n° more than two landing clearances may be issued to successive arrivals. The Safety Board believes that this change will increase system effectiveness, while not creating an undue burden on the controller. Nevertheless, pilots also have a responsibility for separation assurance on the runway and vigilance during landing must be a shared. The Safety Board is aware that if the local controller had provided traffic information to the crew of EA 111, the accident might have been prevented. This procedure, had it been followed, would probably have prompted the crew to query the local controller as to the position of their traffic on the runway, since it was unlikely that visual observation would have occurred. As a_ system redundancy, the Safety Board believes that the importance of issuing traffic information to arriving aircraft should be stressed. Therefore, the Safety Board believes that a mandatory, formal briefing should be provided to all air traffic controllers on the importance of, and the need for, giving traffic information when issuing an anticipated separation landing clearance. The briefing should be contained in an Air Traffic Bulletin. 2.5 Efforts to Reduce Runway Incursions The Safety Board’s concern about the hazard of runway incursions dates back to 1972 following an accident at the Chicago O’Hare International 37 Airport.> As a result of that accident, four Safety Recommendations were issued to the FAA addressing air traffic control procedures and pilotcontroller communications.® The Board’s concerns were further reiterated in 1979 following two more runway incursions incidents and one accident.’ These occurrences prompted the Board to recommend that the FAA conduct a directed safety study to examine the runway incursion problem and to formulate recommended remedial action to reduce the likelihood of such hazardous conflicts. That recommendation was issued in June 1979. In response, the FAA commissioned the Transportation Systems Center in Cambridge, Massachusetts, to conduct a stucy. The study was completed in Apri) 1981 with a report. entitled "An Analysis of Runway-Taxiway Transgressions at Controlled Airports." The study concluded that “there does not appear to be any pattern to the causes . . . other than human errors on the part of both air traffic controllers and pilots." The study also concluded that “more uniform communication and verification of messages between pilots and controllers could serve to reduce the chance of ambiguous or erroneous commands/actions." The report raised the question as to whether system reliability might be improved by increasing the reliability of the human element or by adding redundant elements. The study did not evaluate controller training or human performance issues. The study did suggest that incident reporting might be part of the problem since there were indications that not all incidents are reported, which caused a situation that precluded appropriate corrective measures.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] Pilots must recognize the level of radar services they are receiving. In areas where traffic separation services are not being furnished they must be aware of this, and that they will be required to make a more diligent effort, not only to find conflicting traffic, but to keep previously acquired traffic in sight until they are absolutely certain it is no longer a factor to their flight. These efforts may even require that they maneuver their aircraft in a manner that will enhance their ability to sight and to maintain sight of conflicting traffic. Controllers seem to similarly relax vigilance. The evidence permits an inference that the vigilance of the approach controller and his standards for assessing the resolution of possible conflicts may have lowered because he believed that the flightcrew which had reported traffic "in sight" had a better view of the traffic and a better grasp on the situation than he did. This accident illustrated that this is not a hard and fast rule on which the controller can rely. Even though ~ 35 - the pilot had assumed the burden of maintaining separation, the controller should have not assumed that the pilot's ability to do so will remain unimpaired. He should be prepared to update the pilot's information, and, time permitting, stand ready to alert the pilot to changes in the situation. The principle of redundancy has been recognized as one of the foundations of flight safety, and redundancy between the pilot and controller can only be achieved when both parties exercise their individual responsibilities fully regardless of who has assumed or been assigned the procedural or regulatory burden. 3. CONCLUSIONS XN 3.1 Findings 1. Flight 182 was cleared for a visual approach to runway 27 at Lindbergh Field. 2. The Cessna was operating in an area where ATC control was §, being exercised and its pilot was required either to , comply with the ATC instruction to maintain the 070° heading or to advise the controller if he was unable to do so. 3. The Cessna pilot failed to maintain the assigned heading contained in his ATC instruction. ; 4, The cockpit visibility study shows that if the eyes of the Boeing 727 pilot were located at the aircraft's design eye reference point, the Cessna's target would have been visible. 5. Two separate air traffic control facilities were controlling traffic in the same airspace. 6. The approach controller did not instruct Flight 182 to maintain 4,000 ft until clear of the Montgomery Field airport traffic area in accordance with established procedures contained in Miramar Order NKY.206G. 7. The issuance and acceptance of the maintain-visual-~separation clearance made the flightcrew of Flight 182 responsible for seeing and avoiding the Cessna. - 8. The flightcrew of Flight 182 lost sight of the Cessna and did not clearly inform controller personnel of that fact. 9, The tower local controller advised Flight 182 that a Cessna was at 12 o'clock, 1 mile. The flightcrew comments to the local controller indicated to him that they had passed or were passing the Cessna. 10.

[PROBABLE CAUSE] Although the majority has now added as a conclusion, "Two separate air traffic control facilities were controlling traffic in the same airspace,'' there is no discussion in the ~ report to support this conclusion. Such a procedure is not the most efficient or the safest way to handle traffic; it would have been far better if only one facility was handling both ‘aircraft, since the communications to both aircraft would then have been much more expeditious, meaningful, and efficient. The lack of coordination was emphasized by the mishandling of the conflict alert. Contrary to the majority, I would cite the improper resolution by the controller of the conflict alert as contributory. The Air Traffic Control handbook, 7110.65A, requires a controller to resolve all conflict alerts. The controller failed to do this. The conflict alert was received approximately 19 seconds before the collision. Although this _, might be considered a rather short time, it was still sufficient to have permitted the controller to relay this information to either the Cessna or to the Lindbergh Tower or to have attempted to relay it. Irrespective of the time element, the controllers had no knowledge that there were only 19 seconds to collision, but the duty still existed. According to the majority, the reason the controllers did not take the required action was they considered that the conflict had been resolved based upon PSA 182's response to the traffic advisory, "Traffic in sight." This response had been made 66 seconds prior to the conflict alert and, in my opinion, the controller should not have assumed in such an area as San Diego that the situation was static and that the conflict was resolved. I am at a loss to understand the reasons the majority did not include this failure as a contributing factor .since it is stated in the report (p. 31), "...the failure of the procedures /conflict alert/ to require this to be done may have deprived the pilots of one more chance to avoid the collision." The existing procedures did require action to resolve the conflict. The issuance of a previous visual separation clearance by no means resolves a later conflict. The majority has now concluded that the Cessna failed to maintain the assigned heading contained in the ATC instruction, but it is not cited as a contributing factor for some unknown reason. In my opinion, the failure of the Cessna to maintain the assigned and. mandatory heading was a critical factor in this accident. If the required heading had been maintained, the aircraft would have been separated 1,000 feet vertically; therefore, it is a factor to be considered as contributory. The Cessna was told to "maintain a heading of 070 and vector final approach," which was a mandatory instruction to maintain a heading until the controller was able to vector the aircraft to a downwind leg and the final approach course. This procedure was obviously for separation reasons, since the Cessna was crossing and ascending toward the flightpath of the descending PSA 182. However, the Cessna turned to a downwind leg of 090 prematurely and beneath PSA 182. If this had not been done, the accident May not have occurred.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] The failure to notify controller personnel specifically that they had lost sight of the traffic could indicate that they were not aware of what was embodied in the instruction and that they may have considered it as merely another traffic advisory. The company's chief pilot testified that the procedures embodied in the visual separation clearance are set forth in the regulations, which his pilots carry with them on all flights. He further testified that they are well aware of the requirements embodied within the instruction. However, the visual separation procedures are contained in the AIM and not in the Federal regulations carried by the pilots. He -~ 34 - He stated that AIM information is excerpted for presentation to their flightcrews in ground school, but he could not identify precisely what areas of information were used. The evidence indicates that there may be a communications gap between pilots and controllers as to the proper use of the ATC system. The ATC controllers are responsible for, and are required to apply, the procedures. contained in Handbook 7110.65A in their control of traffic. Despite the fact that the successful use of these procedures requires a mutual understanding on the parts of pilots and controllers of the other's responsibilities, pilots are not required .to read Handbook 7110,65A. One Federal publication containing a description of the interrelationship of pilot and controller roles and responsibilities is the AIM, and this is not--by regulation--required reading for pilots. Considering the responsibilities placed on both the pilot and the controller for the safe operation in the National Air Space system, industry and the Federal Aviation Administration must take steps to insure that the pilots are made cognizant of what this relationship requires of them. Either the AIM should be compulsory reading for all pilots--at least those sections relating to ATC rules, procedures, and pilot and controller roles and responsibility--or pilots should be tested annually or semiannually on their knowledge of these procedures. In conclusion, the evidence indicates that even though flightcrews are still in a "see and avoid" environment, they exercise a lower degree of vigilance in areas where they receive radar assistance than in non-radar areas. Instead of attempting to seek, acquire, and then maintain visual contact with traffic, they seem to rely on the radar and radar controller to point out the aircraft, particularly an aircraft that may be in conflict with theirs. Pilots also seem to have a lessthan-complete knowledge of the specific type of traffic separation services being provided. The types of traffic’separation procedures available in a TRSA vary from that provided in a Stage II and Stage I area, At San Diego, depending either on the aircraft's position or altitude, or both, the pilots could receive either Stage II or Stage III services and could pass rapidly from one area to another. Pilots must recognize the level of radar services they are receiving. In areas where traffic separation services are not being furnished they must be aware of this, and that they will be required to make a more diligent effort, not only to find conflicting traffic, but to keep previously acquired traffic in sight until they are absolutely certain it is no longer a factor to their flight.

[CONCLUSIONS > FINDINGS] The flightcrew comments to the local controller indicated to him that they had passed or were passing the Cessna. 10. The traffic advisories issued to Flight 182 by the approach controller at 0900:15 and by the local controller at 0900:38 did not meet all the requirements of paragraph 511 of Handbook 7110.65A, 11. The approach controller received a conflict alert on Flight 182 and the Cessna at 0901:28. The conflict warning alerts the controller to the possibility that, under certain conditions, less than required separation may result if action is not, or has not been, taken to resolve the conflict. The approach controller took no action upon receipt of the conflict alert, because he believed that Flight 182 had the Cessna in sight and the conflict was resolved. 12. The conflict alert procedures in effect at the time of the accident did not require that the controller warn the pilots of the aircraft involved in the conflict situation. 13. Both aircraft were receiving Stage II terminal radar services. Flight 182 was an IFR aircraft; the Cessna was a participating VFR aircraft. Proper Stage II services were afforded both aircraft. 14. Stage II terminal service does not require that either lateral or vertical traffic separation minima be applied between IFR and participating VFR aircraft; however, the capability existed to provide this type separation to Flight 182. 15. The Boeing 727 probably was not controllable after the collision. 3.2 Probable Cause The National Transportation Safety Board determines that the myle cause of the accident was the failure of the flightcrew of = @tal radar separation to either aircraft. Recount AMENDED, nl 53: - 4, SAFETY RECOMMENDATIONS As a result of this accident, the National Transportation Safety Board has recommended that the Federal Aviation Administration: "Implement a Terminal Radar Service Area (TRSA) at Lindbergh Airport, San Diego, California. (Class I-Urgent Action) (A-78-77)" "Review procedures at all airports which are used regularly by air carrier and general aviation aircraft to determine which other areas require either a terminal control area or a terminal control radar service area and establish the appropriate one. (Class II-Priority Action) (A-78-78)" "Use visual separation in terminal control areas and terminal radar service areas only when a pilot requests it, except for sequencing on the final approach with radar monitoring. (Class 1, Urgent Action) (A-78-82)" "Re-evaluate its policy with regard to the use . of visual separation in other terminal areas. (Class II, Priority Action) (A-78~83)" ERRATA AIRCRAFT ACCIDENT REPORT - Pacific Southwest Airlines, Ine., B-727, and a Gibbs Flite Center, Inc., Cessna 172, N7711G, San Diego, California, September 25, 1978. ; During its evaluation of the ALPA Petition for Reconsideration of Probable Cause of the subject accident, the National Transportation Safety Board also reviewed the entire accident report and its supporting evidence.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] In retrospect, there is little doubt that the controllers were misled (1) by their belief that Flight 182's flightcrew were visually separating their aircraft from the Cessna and (2) by their previous experiences with similar conflict alerts wherein no action on their part was necessary. Based on the procedures, their requirements were satisfied. They, therefore, did not try to reposition and unscramble the data blocks and reacquire the altitude readouts to further monitor the situation because they believed that visual separation was being applied. The Safety Board was not able to determine why Flight 182's and the Cessna's data blocks did not separate automatically. While it was possible that the auto-offset function was enabled at the display but was being delayed by higher priority computer functions, the more likely probability was that the function was inhibited at the display,’ either by the controllers on duty or by controller teams that had worked the display during earlier duty shifts. However, the failure of the air traffic control procedures to require that the controllers notify the pilots that their aircraft were involved in a conflict alert resulted in a less-than-optimum use of the system, particularly in a situation where visual separation procedures were being used in a terminal area. Had this requirement existed, it was possible that warnings and perhaps suggested evasive maneuvers could have been delivered to the pilots of one or even both aircraft. While the Safety Board cannot conclude that the delivery of a warning or suggested instruction to the pilots would have altered the course of events, the failure of the procedures to require this to be done may have deprived the pilots of one more chance to avoid the collision. The planes collided shortly after the tower's traffic advisory. The damage to the Cessna's propeller and matching damage noted on the No. 5 leading edge flap actuator of Flight 182 show that the impact occurred on the forward and underside of its right wing about 12.5 ft outboard of the wing root. Almost every witness who saw the collision confirmed this conclusion. The study of the two photographs showed that the structural damage to the Boeing 727's right wing leading edge extended from the No. 4 inboard leading edge flap outboard to, and including, the No, 3 leading edge slat--a distance of 30 feet or more. The chordwise penetration of this damage appeared to extend rearward to the front spar of the wing. The calculated positions of the flight controls in Figure 2 show almost full deflection in the proper direction to arrest the abnormal attitude and to restore controlled flight. The deflected position of the flight controls and the left wing flight spoiler surfaces indicated that at least partial hydraulic pressure was available from system A and system B. -~ 32 - The Safety Board was not able to assess precisely what effect the structural damage, the impingement of Cessna parts on the structure, and the existing fire had upon Flight 182 aerodynamic capabilities and control effectiveness. Considering the extent and magnitude of the collision damage, the Safety Board concludes that the aircraft was probably uncontrollable.

[PROBABLE CAUSE] At 0901:38 and 0901:39,.the first officer pointed out a target, "There's one underneath," and "I was looking at that inbound there." The only known and reported inbound traffic was a PSA flight that at this time had completed its landing roll and was in the 6 o'clock position to PSA 182. The first officer could not have been looking at this aircraft but must have been looking at unreported and unknown inbound _ traffic. Significantly, 16 ground witnesses reported additional traffic in the area that could be interpreted as being votential traffic to PSA 182. However, the important fact is, there appears to have been at least one inbound aircraft that was unknown or unreported by ATC. . Despite the conclusion of the majority that the evidence indicates there was not a third aircraft in the area, my reading of the evidence is contrary. The evidence is inconclusive on this point, and the existence of a third unknown or unreported aircraft was a distinct possibility. If there was a third aircraft and the crew of PSA 182 was watching it, this could explain the reason why the crew of PSA 182 either did not see the Cessna or subsequently lost contact with it. , Based upon the foregoing, I would state the probable cause as follows: "...was the failure of the flightcrew of Flight 182 to maintain visual separation and to advise the controller when visual contact was lost; and the air traffic control procedures in effect which authorized the controllers to use visual separation procedures in a terminal area environment when the capability was available to provide either lateral or vertical radar separation to either aircraft. Contributing to the accident were: 1. The failure of the air traffic control system to establish procedures for the most effective use of the conflict alert system at the San Diego approach control facility. - 44 ~ The failure of the controller to restrict PSA 182 "to a 4,000-foot altitude until clear of the Montgomery Field airport traffic area. The .improper resolution by the controller of the conflict alert. The procedure whereby two separate air traffic control :facilities were controlling traffic in the same airspace. The failure of the controller to advise PSA 182. of the direction of movement of the Cessna. The failure of the Cessna to maintain the -assigned heading. The possible misidentification of the Cessna by PSA 182 due to the presence of a third unknown aircraft in the area. /s/__ FRANCIS H. McADAMS Member ~- 45 - APPENDIX A Investigation and Hearing 1. Investigation The National Transportation Safety Board was notified of the accident about 1210 e.s.t. on September 25, 1978, and immediately dispatched an investigative team to the scene.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] Considering the extent and magnitude of the collision damage, the Safety Board concludes that the aircraft was probably uncontrollable. Although the evidence showed that approved ATC separation procedures were used by the controllers, the Safety Board's investigation disclosed other areas which may have contributed to the accident, Although Flight 182 was provided all the services appropriate under Stage II radar procedures, these procedures merely helped the pilot apply the regulatory "see and avoid" principles. The Safety Board recognizes that some level of "see and avoid'' will remain a valid concept for collision avoidance whenever an aircraft is flown in visual conditions and will be a part of any collision avoidance system. However, the . concept appears to place a disproportionate burden on the flightcrews of “air carrier aircraft, high performance general aviation aircraft, and high performance military aircraft. This is especially true where the concept is used for collision avoidance in a mixture of high-speed and low-speed traffic in a terminal area. Because of the performance characteristics of their aircraft, these flightcrews are almost always operating the overtaking aircraft, and, therefore, are solely responsible for avoiding the slower moving aircraft. Their overtake rate is usually high, and they can expect little assistance from the other aircraft. Since most of these aircraft are flown by two or more persons, one might conclude that the avoidance problem would be lessened. However, several factors reduce the amount of time spent in traffic scan. Configuring these aircraft for landing requires the execution of a checklist, and many of these checklist items require attention after the aircraft has entered the terminal traffic mix. Many of these aircraft require several flap settings and airspeed adjustments to reach the landing flap configuration. These aircraft generally enter the > terminal area on a descending flightpath that ends either at entry into the traffic pattern or at the beginning of the final approach, These descents are often flown with the aircraft in a noseup deck angle, which limits the flightcrew's visibility in the area where they are descending. Finally, the traffic they are required to detect and avoid may not be detected easily and may be further camouflaged by the surface background. While extra persons may aid in the scan, the pilot must manage his cockpit to insure that the extra person either assists in the scan, or does not interfere with it. In this instance, although the captain and first officer saw the aircraft, there is no evidence to indicate that it was pointed out to any other cockpit occupant. Although company procedures urge the flight engineer to plan "routine paperwork and radio contacts . . . to be accomplished at altitudes above 10,000 ft," he was involved with radio contacts with the company when the Cessna was pointed out to Flight 182 and the visual separation instruction was issued.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] In this instance, although the captain and first officer saw the aircraft, there is no evidence to indicate that it was pointed out to any other cockpit occupant. Although company procedures urge the flight engineer to plan "routine paperwork and radio contacts . . . to be accomplished at altitudes above 10,000 ft," he was involved with radio contacts with the company when the Cessna was pointed out to Flight 182 and the visual separation instruction was issued. Since the extraneous conversation within the cockpit ceased after the flightcrew told the approach controller that they had the Cessna in sight, the conversation cannot be considered a contributing factor. ‘However, this conversation persisted until the flight descended to 3,200 ft and while a checklist was being accomplished. Even though a flightcrew is responsible primarily for communications addressed to them, advisories to other aircraft can be valuable and may aid in their assessment of traffic which could become a factor. According to the CVR, at 0857:44, while the extraneous conversation was in progress, a company flight preceding Flight 182 was advised of the presence of the Cessna and its future flightpath. The first officer asked if the message, which included a clearance to the tower frequency, was for Flight 182. Since the message was not for Flight 182, no assumption can be made as to whether or not its flightcrew heard or understood the advisory preceding the clearance. Although the conversation was not causal, it does point out the dangers inherent in this type of cockpit environment during descent and approach to landing. The issuance of the "maintain-visual-separation" clearance and Flight 182's response to the instruction raises several areas of concern. This method of separation can be applied not only in Stage II, but also in a TRSA and a Terminal Control Area. The use of this type of separation does little else but place the pilot into a "see and avoid" situation even though he is flying in an area where the ATC system is capable of providing vertical or lateral separation. San Diego approach control had the capability of providing either vertical or lateral separation criteria between IFR aircraft and participating VFR aircraft. Had this been done, Flight 182 and the Cessna would not have collided. The Safety Board believes that participating aircraft operating on random courses to each other should be afforded this type of separation until they are clear of each other. This would be particularly appropriate for high performance aircraft. Based on available evidence, the Safety Board cannot conclude whether the flightcrew of Flight 182 knew what they were required to do when they accepted the "maintain-visual-separation" clearance from the controller. In addition to maintaining proper separation from the designated aircraft, their acceptance of the clearance required them to tell the controller when they. no longer had it in sight. The failure to notify controller personnel specifically that they had lost sight of the traffic could indicate that they were not aware of what was embodied in the instruction and that they may have considered it as merely another traffic advisory. The company's chief pilot testified that the procedures embodied in the visual separation clearance are set forth in the regulations, which his pilots carry with them on all flights.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] The controller stated that this was a primary radar return, that it had passed Flight 182, and that he had i no idea of its altitude or where it went after that. At 0859:39, the approach controller advised Flight 182 of "additional traffic" and described the aircraft type, location, heading, and altitude. The advisory described the Cessna's heading and its position in relation to Lindbergh Field. At 0859:50, the first officer told the approach controller that "Okay, we've got that other twelve." At 0900:15, the approach controller again advised Flight 182 of the Cessna's position and altitude. Since this traffic advisory did not contain the direction of traffic movement or the aircraft type, it did not meet the requirements of Handbook 7110.65A, paragraph 511. However, at 0900:21, the first officer said, "Got em", and 1 sec later the captain told the controller, "Traffic in sight." The approach controller cleared the flight to maintain visual separation and to contact the Lindbergh tower, and the captain answered, "Okay." The acceptance of a "maintain-visual-separation" clearance requires that the pilot separate his aircraft from traffic that has been pointed out to him. While there was no doubt that the controller was pointing out the Cessna to the crew of Flight 182, the question arises as to whether the flightcrew was referring to it when they called "traffic in sight." The two traffic advisories concerning the Cessna placed it at 1,400 to 1,700 ft, northeastbound, just north of Lindbergh Field, and in front of Flight 182. If the flightcrew had identified another aircraft as the Cessna at this time, then it is logical to assume that it was flying in the same area about the same time as the Cessna and on a similar course and altitude. In order to be flying in this area it would have had to have been operating within Lindbergh Field airport traffic area. About the time of the collision Lindbergh tower controllers were in radio contact with two airborne aircraft-~-Flight 207, a Boeing 727 which took off at 0901:47, and a Cessna 401, N3208Q which was 9.5 nmi east of the field. Therefore, if a third aircraft was operating in this area its pilot was doing so in violation of Federal regulations. All of the witnesses who saw another aircraft in the vicinity saw it either immediately before, during, or just after the collision; however, no one saw another small aircraft just north of Lindbergh and on a northeasterly track at the time the Cessna was in the area. Thus, it was necessary to determine if any of these aircraft could have transited the area north of Lindbergh at the time the Cessna was sighted by the flightcrew of Flight 182. It was highly improbable that there were 16 different small aircraft in the area during the time interval described above; however, there was no one aircraft track that was supported by a majority of the witnesses.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] The reason for the Cessna's deviation from the heading could not be determined; however, the pilot was flying in an area in which air traffic control was being exercised and he either should have complied with the instruction or informed the controller otherwise. At 0900:31, the controller informed the pilot of the Cessna of the presence of Flight, 182. This advisory was given while N7711G was still on what appeared to be a crossing track to that of Flight 182. Shortly thereafter, the Cessna began a right turn to a flightpath that would coincide with Flight 182's flightpath. According to the visibility study, during the time between this advisory and the collision, Flight 182 would not have been visible to the Cessna pilots. Since the Cessna pilots were told that they were being overtaken by an aircraft whose flightcrew had them in sight, it would be unrealistic to conclude that they would have made any attempt to turn their aircraft in order to sight Flight 182, Regardless of the Cessna's change of course, Flight 182 was the overtaking aircraft and its flightcrew had the responsibility of complying with the regulatory requirement to pass "well clear" of the Cessna. The regulations do not establish minimum lateral and vertical separation distances for this maneuver; consequently, the 'well clear" distance was a matter of pilot judgment, and, as stated by the company's chief pilot, 1/2 mile would have been adequate separation for this maneuver, even though it would place the aircraft within the conflict alert system's Type III warning parameters. The conflict alert warning began about 19 sec before the collision. Handbook 7110,65A required a controller to take appropriate action to resolve a conflict when the alert is displayed; however, he must also decide if the conflict has been resolved. Corrective actions do not necessarily require the controller to notify a pilot that his aircraft is involved in a conflict. For example, in this case, the ' responsibility for separation was in the cockpit of Flight 182, and while the separation maintained by that flightcrew did not satisfy the conflict alert computer, it could have been more than adequate for clearing the Cessna in visual flight conditions. The approach controller's decision of whether this conflict had been resolved or whether it required action on his part was based on his judgment and experience. Based on all information available to him, he decided that the flightcrew of Flight 182 were complying with their visual separation clearance; that they were accomplishing an overtake maneuver within the separation parameters of the conflict alert computer; and that, therefore, no conflict existed. In retrospect, there is little doubt that the controllers were misled (1) by their belief that Flight 182's flightcrew were visually separating their aircraft from the Cessna and (2) by their previous experiences with similar conflict alerts wherein no action on their part was necessary.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] Contrary to Miramar Order NKY 206G, the approach controller at San Diego approach control did not direct Flight 182 to maintain 4,000 ft until clear of the Montgomery Field airport traffic area. The controller said that Flight 182 was outside the area when he cleared it for the visual approach and that he monitored its course on his radar. Since the flight did not enter the Montgomery Field airport traffic area, he said there was no need either to place the restriction on the flight or coordinate its passage with Montgomery Field. His determination was based on the fact that Flight 182's course placed it south of the MZB VORTAC's 090° radial which, to him, constituted the end of the Montgomery Field airport traffic area and the beginning of the Lindbergh Field air traffic area. However, Flight 182's ground track showed that it passed about .8 mile inside Montgomery Field's airport traffic area. The purpose of the altitude restriction in the order was to avoid a potential conflict with Montgomery Field operations. In this instance neither aircraft was a Montgomery Field operation. One could infer that, had the restriction been applied to Flight 182, the two aircraft would have remained separated and that, even though the Cessna was not a traffic operation protected by the order, the failure to apply it was a causal factor. This inference might be valid if the controllers had taken no other action to insure that they were separated; however, they did take other action. The evidence is conclusive that the controllers pointed out the traffic to Flight 182 and then applied approved traffic separation procedures to geparate the aircraft. After Flight 182 was cleared for the visual approach, it was I still an IFR flight although it was operating in visual flight conditions! Federal regulations required the crew to ''see and avoid" other aircraft. Stage II radar services are designed to aid the pilot in accomplishing this regulatory responsibility. Thus, beginning at 0859:30, Flight 182 was given three traffic advisories by the approach controller, and one by the Lindbergh tower local controller. At 0859:30, Flight 182 was advised of traffic 1 mile in front of it and heading in a northerly direction. The crew's response indicated that they did not see the aircraft and were looking for it. The controller stated that this was a primary radar return, that it had passed Flight 182, and that he had i no idea of its altitude or where it went after that. At 0859:39, the approach controller advised Flight 182 of "additional traffic" and described the aircraft type, location, heading, and altitude. The advisory described the Cessna's heading and its position in relation to Lindbergh Field.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] While it is possible that all this might have occurred, the weight of the evidence indicated that there was not a third aircraft in the vicinity of the Cessna that could have been mistaken for it by the flightcrew of Flight 182. The visibility study showed that when the 0859:39 and 0900:15 advisories were issued, the Cessna would have been almost centered on both pilots' windshields. Even if their eyes were lower and slightly aft of the design eye reference points, the cockpit structure of the~ Boeing 727 would not have prevented either pilot from sighting the Cessna. Since the sun was above the horizon and the Cessna was below _ it, the pilots would not have had to look directly into the sun to find the Cessna, and the white surface of the Cessna’s wing could have presented a relatively bright target in the sunlight. The cockpit conversation from 0900:15 and 0901:21 showed that the captain and first officer sighted an aircraft; that they had identified the aircraft as a Cessna; that they sighted the aircraft in the same area that the controller had said the Cessna was flying; and that air traffic control was informed that the traffic was "in sight." The evidence showed that the captain and first officer did have the Cessna, N7711G, in view at or shortly after it was first pointed out to them. ~ 29 ~ The evidence was conclusive that the flightcrew's transmissions to the approach controller convinced him that they had the Cessna in sight and that they were capable of meeting the criteria imposed upon them by their acceptance of the instruction to maintain visual separation. The two later advisories issued to N7711G which stated that a "PSA jet" descending into Lindbergh "has you in sight'' offered confirmation of the approach controller's state of mind. From the time he accepted control of Flight 182 until he transferred communications to the tower, the approach controller used the procedures prescribed by Handbook’ 7110.65A, with the one exception noted earlier. At 0900:38, Flight 182 received its last traffic advisory. The Lindbergh tower local controller advised that there was a Cessna 1 mile in front of the flight. The advisory was based on the portrayal of the BRITE 4 radar display and it was timely. Although the controller did not scan the area visually, it is doubtful that values and directions derived from the radar display could have been improved upon by an estimate based on visual observations of two aircraft that were over 2 mi from the tower and were separated from each other by at least 1 mi. However, the advisory did not contain the direction of traffic movement; therefore, it did not comply with the provisions of paragraph 511 of the Handbook 7110.65A.

[PROBABLE CAUSE] Therefore, rather than isolated conclusions with little or no support, they should have been cited as contributory. Additionally, as contributing factors, I would have I cited the failure of the controller to restrict PSA 182 to a 4,000-foot altitude until clear of the Montgomery Field airport traffic area. The evidence is clear that PSA 182 was approximately eight-tenths of a mile inside the Montgomery Field traffic area and therefore should have been restricted to the 4,000-foot altitude. The majority eliminates this issue as a contributing factor for the reason the controller took other action to insure the separation of the aircraft. The other action was to issue a visual separation clearance. This action,of course, is not relevant, since the imposition of the restriction does not depend upon other action; it is to be imposed in all cases upon southbound air carrier aircraft into the San Diego area. If the restriction had been imposed, the accident possibly would not have occurred, and therefore, it should be considered as contributory. I would also assign as a contributing factor the failure of the controller to advise PSA 182 of the direction of movement of the Cessna. The last two traffic advisories at 0900:15 and 0900:38 eliminated the direction of movement of the Cessna. I believe this to be a critical omission since it is not only required but is an essential aid to the pilot in acquiring and maintaining the traffic that has been pointed out. If the crew of PSA 182 had known the direction of movement, it is possible the target would not have been lost. Also, if these advisories had contained the direction of movement and PSA had replied "traffic in sight," the possibility of misidentification or any misunderstanding would have been substantially lessened, Furthermore, at the time of the second advisory the Cessna had alreddy turned from a heading of 070 to a heading of 090, the same heading as PSA 182. At this time, according to the CVR and the ATC transcriptions, PSA 182 had lost contact with the Cessna, the reason being, obviously, the Cessna had turned beneath PSA 182 and to the same heading. PSA 182 was never advised by ATC that the Cessna which had been previously reported to be on a northeast heading had turned to 090. Therefore, if PSA 182 had been advised that the Cessna was now on a heading of 090 and beneath them, they possibly would have been able to reacquire the I target visually or to request avector for separation. Although the majority has now added as a conclusion, "Two separate air traffic control facilities were controlling traffic in the same airspace,'' there is no discussion in the ~ report to support this conclusion. Such a procedure is not the most efficient or the safest way to handle traffic; it would have been far better if only one facility was handling both ‘aircraft, since the communications to both aircraft would then have been much more expeditious, meaningful, and efficient.

[PROBABLE CAUSE] This procedure was obviously for separation reasons, since the Cessna was crossing and ascending toward the flightpath of the descending PSA 182. However, the Cessna turned to a downwind leg of 090 prematurely and beneath PSA 182. If this had not been done, the accident May not have occurred. In my opinion there still exists the possibility that there was a third unknown and unreported aircraft in the area which could have been mistaken by the crew of PSA 182 for the Cessna. Analysis of the CVR could be interpreted to mean that PSA never acquired the Cessna but was observing some. other aircraft that was unknown or unseen by ATC. Even the majority concedes this point since they state (p. 26), ".,,the question arises as to whether the flightcrew was referring to it /the Cessna/7 when they called ‘traffic in sight.'"' At 0859:39, a traffic advisory indicated the Cessna at 3 miles, and at 0859:50 PSA replied, 'We've got that other twelve.'’ Whether he was referring to a previous traffic advisory or to the Cessna is not clear. At 0900:15 -- 37 seconds after the first traffic advisory -- another advisory was given but without aircraft identification or direction of movement, but still reporting the target at 3 miles. This mileage was corrected at the hearing, but insofar as PSA was concerned these two traffic advisories could have been related to two different aircraft since the second advisory did not either identify the target or the direction of movement, and the distance remained the same, 3 miles. Obviously, the mileage would have changed by approximately 2 miles between the two aircraft, and at the time of the second advisory the separation was approximately 1 mile. This could have led PSA to assume there were two different aircraft. Further, if PSA 182 had the Cessna in sight at 0900:21 on a north-northeast course, he would have expected the target to pass off to the left of his aircraft and not to the right as he stated at 0900:50. ae . {{i Additionally, the captain reported he had seen the target at 1 o'clock before turning downwind, whereas it has been well established by the ground track of both aircraft that at this time the Cessna would have been at the 11 o'clock position. This is a difference of approximately 60 degrees, a substantial change, and could indicate the captain was looking at a target other than the Cessna, either unreported or unknown to ATC. At 0901:38 and 0901:39,.the first officer pointed out a target, "There's one underneath," and "I was looking at that inbound there." The only known and reported inbound traffic was a PSA flight that at this time had completed its landing roll and was in the 6 o'clock position to PSA 182.

[PROBABLE CAUSE] A contributing factor is not a primary cause; it is more remote and does not carry the same weight or implications as that of a probable cause. In my opinion, these inadequacies should have been given equal weight in the probable cause with the failure of the PSA crew to maintain visual separation rather than being merely mentioned as a contributory factor. The San Diego approach control had the capability of providing either vertical or lateral separation between IFR aircraft and participating VFR aircraft, and this procedure should have been used for the control of both aircraft. If it had, the accident would not have occurred. Apparently the majority agrees but is either reluctant or diffident to include this issue in the probable cause, since it is stated (p. 33) that if either vertical or lateral separation had been used, ..Flight 182 and the Cessna would not have collided." Such language clearly implies that this omission was a direct cause of the accident and therefore should have been included as a probable cause. The controller, instead of using available procedures, gave PSA 182 a visual separation clearance which placed the pilot in an exclusively see-and-avoid situation where the last redundancy of the system was removed. The redundancy should not have been eliminated in a dense terminal traffic area such as San Diego. In my opinion, the concept of see and avoid is outmoded and should not be used in high volume terminal areas. Positive radar separation should be used with the backup, or redundancy, being the pilot's visual ability to see and avoid. In this case, both aircraft should have remained under positive radar separation since it was available and could have provided safe separation. The failure to do $2, therefore, must be considered as causal. Furthermore, despite strong urging on my part, the majority has not named several other factors which I consider as being contributory. It is true that the majority has included three issues which I had suggested as contributing factors, but they have been included in the report only as conclusions. For example, the majority concludes that the approach controller failed to restrict Flight 182 to a 4,000 - foot altitude; obviously, that logically means the controller had a duty to issue an altitude restriction, and if such altitude restriction had been issued, it is possible the accident would not have occurred, Ergo, it is a contributing factor as well as a conclusion. A similar argument can be made with respect to the other two conclusions of the majority, i.e., the Cessna failed to maintain the assigned heading, and two separate facilities were controlling traffic in the same airspace. Therefore, rather than isolated conclusions with little or no support, they should have been cited as contributory. Additionally, as contributing factors, I would have I cited the failure of the controller to restrict PSA 182 to a 4,000-foot altitude until clear of the Montgomery Field airport traffic area.

[ANALYSIS] Terminal Airspace," Billings, Grayson, Hecht, and Curry, NASA TM 81225, August, 1980. ~22- cannot inform the pilot of traffic that is not visible on his radar scope, nor can he provide separetion from such traffic. It is plain that at least some pilots receiving Stage III services believe that they will be told about all traffic that represents a threat, vet controllers can handle traffic only with regard to threats they can see...” A general aviation pilot in one case study said: ...1 have been able to practice more effective collision avoidance by listening to communications on the frequency than by receiving advisories... I'm afraid many pilots get a false sense of security when under radar control or advisory. ... Those pilots who do not understaid them must be taught the limitations of terminal radar, and ef the controllers who use it as their primary source of information. Many aircraft in TRSA's, and some intruders in TCA's, are not transponder-equipped; such aircraft are often not visible to controllers. These aircraft, and many others near TCA boundavies, may represent 4 threat detectable only by the pilot, and then only if he is looking for them. The highest level of pilot vigilance must be mainta:ned to avoid midair collisions, regardless of the airspace in which operations are being conducted and regardless of the ATC services being utilized. No pilot should permit himself to be lulled into a false sense of security by ATC procedures that cannot necessarily guarantee separation under visual meteorological conditions.... The system of separation assurance is not 'error-proof,' nor, in all probability, will it ever be. Separation can be assured most effectively by providing air traffie controllers with the best possible information about all aircraft within their area of responsibil‘ty; by minimizing flighterew workload in terminal airspace, thus permitting them to maintain the best possible outside surveillance; and by making pilots aware of the critical importance of maintaining such surveillance, regardless of the services they are receiving. It is hoped that this study and report will help to increase that level of awareness.... In summary, the authors of the 1980 NASA study concluded that: "A variety of human and system factors was found to be asscciated with these near midair eollisions. Flighterew workload, limited visual scan while under racar control, misunderstanding of the lirnitations of the ATC system, and failure to util’ze transponders were observed. A substantial number of reported near midair collisions in Stage II terminal airspace involved at least one aircraft not participating in Stage Ill services.

[ANALYSIS] The pilot testified that he was about to call the Denver Center when the accident occurred. Since he was at 13,000 feet and ~17- climbing to 15,500 feet at a rate of about 500 feet per minute, he would have been in compliance with the 5-minute jumping advisory provision of FAR 105.14. However, if the pilot had complied with FAR 91.24 and requested an ATC authorized deviation he would have had to communicate earlier with the Denver Center. This communication would have alerted the controller to the presence of the Cessna in his sector, and if he granted the deviation and radar identified the aircraft, he could have started to "track" the target. Thus, the Safety Board believes that the pilot's failure to communicate with the Denver Center forestalled the use of any possibile air traffie control procedures which could have been used by the controller to track this aircraft and which possibly could have averted the collision. It should also be noted that the pilot :” the Cessna had frequently been assigned code 1-2-3-4 by the controllers for use auring parachute jump operations. Consequently, the pilot believed that this was a permanently assigned code and that by merely squawking 1-2-2-4 positive radar identification was provided for the aircraft. This misconception created an unsafe condition in that it provided a <alse sense of security for the Cessna pilot. The Sector 14 controller did not recall having seen a target associated with the Cessna and observed no conflicting traffic for Air U.S. 716. The non-Mode-C filter button was not activated thereby eliminating radar target display of non-Mode-C equipped aircraft. If th: non-Mode-C filter key had been activated, the radar return of the Cessna would have been continuously visible on the display. Even this would not necessarily have alerted control personnel to a potential conflict since they would have expected all non-Mode-C targets to be below 12,500 feet. However, if the accident aircraft had been equipped with a Mode-C transponder, the controller's radar display would have depicted the aircraft as a specific target inc’uding a data block containing three digit altitude information and a ‘our digit transponder code. This would have provided ample information which the controller could have used for separating the jump aircraft from any conflicting traffic. 2.1.2 Distribution of NOTAM Regarding Jumping Activities The information necessary for the Flight Service Station to issue a NOTAM was provided by Sky's West 23 minutes before their first flight of the day rather than 1 hour before, as required by FAR 105.23. The notification, once received by the Denver FSS, was required to be treated as a NOTAM with local distrioution but was merely posted on the FSS weather board. The FSS team supervisor stated that he believed the NOTAM pertained to the Ft. Collins Yankee jump area (about 10 nmi north of Ft. Collins/Loveland). The Yankee area is listed in the U.S.

[ANALYSIS] While searching a clear sky or a homogeneous field tends to produce a condition in the viewer's eyes known as “empty field myopia" in which the eyes will accommodate or tend to focus at a distance of 30 to 35 feet becaus* no specific reference psints are present, empty field myopia was not a factor in this accident. Tne Jetstream erew would have had several points to focus on, such as the horizon, mountains, and clouds. 2.3.3 Pilot Vigilance The possibility of « pilot's detevting airborue targets depends upon his expectations in finding a target thas he has been alerted to, 1s physical well-being, how he time-shares the instrument scanning and outside scanning, and the techniques used in searching for uirborne targets. Obviously, if a pilot assumes that he is protected by ATC and/or is fatigicd, bored, preoceupied, or distracted, his ability to sean the airspace while simultaneously watching coekpit displays, flying the nireraft, and rnonitoring ATC communications wils be seriously impaired. In this aecident, there was no evidence to indicate that the Jetstream pilots were fatigued or physically unfit. It is not possible to determine how much time during the final 120 seconds of flight each pilot could have devotes io outside scanning, nor is it known what each pilot's seanning habits or techniques might have been. A recent NASA study of data from the Aviati.n Safety Reportine System (ASRS) on near midair collisions i9/ indicated that half of 78 near midair collisiens in Terminal Controlled Airspace (TCA's) involved one aircraft not known to ATC, "if ASRS reports are representative, inany pilots under radar control believe that they will be advised of traffic that represents a potential conflict and behave accordingly. They tend to relax their visual sean for other aircraft until warned of its presence; when warned of a conflicting aircraft, they tend to look for it to the exelusion of within-cockpit tasks end seanning for unreported traffic." The report continues: "The air traffic controller 10/7 "A Study of Near Midair Ccilisions in U.S. Terminal Airspace," Billings, Grayson, Hecht, and Curry, NASA TM 81225, August, 1980. ~22- cannot inform the pilot of traffic that is not visible on his radar scope, nor can he provide separetion from such traffic.

[ANALYSIS] Flighterew workload, limited visual scan while under racar control, misunderstanding of the lirnitations of the ATC system, and failure to util’ze transponders were observed. A substantial number of reported near midair collisions in Stage II terminal airspace involved at least one aircraft not participating in Stage Ill services. For these reasons, pilots must exercise the highest level of vigilance for other traffic, regardless of airspace or radar services being utilized.” Although the Safety Eoard could not determine precisely why the Jetstreani flightcrew did not see the Cessna 206, these conclusions are applicable to the present accident situatior as likely explanations for the failure of the "see and avoid" concept to have prevented this collision. The Safety Board recognizes the inherent limitations of the see and avoid concept and have cited them in numerous Board reports involving midair collisions. AJthough the FAA has published considerable data regarding the need for continued pilot vigilance in order to minimize the collision hazard, the Board believes that there is still insufficient, detailed information available for the enlightenment of pilots and controllers regarding the limitations associated with this concept. Notwithstanding the above cited limitations, the Safety Board believes that strict adherence by al! concerned to existing rules contained in FAR 91 and 105 and applicable procedures set forth in the Airman’s Information Manual ould possibly have prevented this accident. 3. CONCLUSIONS Findings The flighterews of both aircraft were properly certificated and qualified for their flights. The aircraft were certificated and maintained in accordance with applicable regulations. (Except as noted in appendix C.) The pilots of both aircraft were required by regulation to "see and avoid" each other. The pilot of the Cessna TU-266 misunderstood the use of the ATC transponder and based on his prior experience at Denver's Stapleton Airport erroneously, but understandably, interpreted the meaning of the wurd "roger" as an approval by the controller to deviate from the Mode-C transponder requirement above 12,500 feet m..s.1. The pilot of the Cessna TU-206 did not establish and maintain radio contact with Denver Center as required by Sky's West procedures. The Cessna pilot continued flight to an altitude above 12,500 feet m.s.1. without a Mode-C encoding altimeter aboard the aircraft as required by FAR 91.24(bX4) and without authorization to deviate from _ the regulation. The Cessna pilet erroneously assumed that he was protected from collisions with other aircraft by ATC even though he never contacted ATC during the accident flight. Had tie Cessna bee. equipped with Mode-C, the resultant target with an indication of the altitude of the Cessna would have been presented - clearly on the controller's radar display. An untracked, beacon reinforced primary target was presented on the controller's display for about 75 percent of the Cessna's flight path, but was not noted by the controller.

[FINDINGS] Had tie Cessna bee. equipped with Mode-C, the resultant target with an indication of the altitude of the Cessna would have been presented - clearly on the controller's radar display. An untracked, beacon reinforced primary target was presented on the controller's display for about 75 percent of the Cessna's flight path, but was not noted by the controller. FAA management personnel at the Denver Center did not take decisive action when they had knowledge of routine parachute jump operations being conducted by Sky's West above 12,500 feet without Mode-C transponders. The Flight Service team supervisor did not disseminate the NOTAM on the parachuting activity to the Denver Center or to any other facility as required by FAA instructions, Sky's West and FAA did not initiate any action to have the Ft. Collins/Loveland area listed in the NOAA Airport/Facility Directory. Cessrne TU-206 binocular phctographs taken 1 inch above the CAM4b desigr. eye reference point indicate that the Jetstream would have been ~24- within the binocular vision envelope of the pilot's windshield for at least a 45-second interval, beginning 120 seconds before the collision, but not for the last 60 seconds. Any aft movement of the Cessna pilot's seat would have altered his I physical constraints to visibility and reduced the binocular vision envelope. The Cessna pilot was not looking for traffic prior to the collision because he was looking at the airport and drop zone. Jetstream binocular photogranhs taken at the design eye reference point indicate that the Cessna would have been present within the binocular vision envelope of both pilots! windshields for about a 45-second interval begining about 60 to 75 seconds before the collision, The physical constraints to visibility for the Jetstream flighterew would not have been significantly altered by the flightcrew's seat adjustments. The Jetstream crew had not been advised of any traffic in its area and may not have been scanning for traffic in any particular sector just before the collision. Psychophysiologicai factors and Scanning techniques could have affected the Jetstream flightcrew's ability to detect and identify the Cessna as a potential hazard. 3.2 Probeble Cause The National Transportation Safety Board determines that the probable cause of the accident was the failure of the Cessna pilot to establish communications with the Denver Center and his climbing into controlled airspace above 12,500 feet without an authorized deviation from the altitude encoding transponder (Mode-C) requirement, the practice of the Denver Center of routinely condoning Sky's West parachute jump operations above 12,500 feet without a Mode-C transponder and the failure of the pilots of both aircraft to "see and avoid" each other. Contributing to the accident was the fact that existing regulations do not prohibit parachute jumping in, or immediately adjacent to, Federal airways.

[ANALYSIS] A reduced state of vigilance would explain why they failed to detect the presence and conflict presented on their radarseope by the LDB representing the PA~31 airplane. Within the last 12 months the Safety Board has investigated five midair collisions in which the air traffie controller workload was judged light or moderate yet the controllers did not perceive a collision threat and did not issue traffic advisories or safety alerts beforc any of the collisions. The apparent pattern suggests that periods of low air traffic controller workload may result in periods of reduced vigilance on the part of the controllers and procluce a greater hazard to traffic separation than had been previously recognized. li the Safety Board's runway incursion special investigation, 14/ it was found that heavy traffic and reduced visibility were infrequently involved. On the contrary, traffic was reported as light or moderate at the time of most of the incursions where controller actions were involved. In some of the controller-induced runway incursions, the controllers were working as few as two airplanes. The FAA (Civil Aeromedical Institute, in a study of ATC operational error incidents occurring from 1965 to 1980, noted <hat 40 to 50 percent of the errors occurred under moderate controller workloads, Over the period evaluated, there was a reported trend toward Increased numbers of incidents occurring during light traffic, 15/ The Safety Bourd believes that it is more likely that the Kansas City TRACON Rast Radar controllers were distracted from monitoring traffic in the moments before the collision because of their position rellef briefing and associated duties than that they were inattentive and not vigilant as a consequence of their otherwise light workload. Nonetheless, the Sefety Board is concerned with the apparent increase of ATC operational errors, runway incursions, and midair collisions which have occurred during periods of low alr traffie controller workload. The Safety Board believes that controllers have a tendency to relax their vigilance in the low workload environment making them susceptible to operational errors and omissions, Further, the Safety Board believes that FAA action Is needed to preclude reduced controller vigilance during periods of low eontroller workload. The Safety Board is aware that both of the Kansas City TRACON Vast Radar controllers and all other controllers employed by the FAA, since the advent of ARTS tracking systeins in the 1970s, were trained to identify and track targets using the ARTS. Recently, the Safety Board has learned that some FAA facilities no longer permit radar controllers to control traffie except by use of the ARTS equipment. The Safety Board is concerned that as a result of their training and possibly operationa) experience, some radar air traffic controllers may be focusing an inordinate amount of attention to targets Identified by FDB targets (tracked targets) to the exclusion of untracked targets identified by LDB or by primary or secondary radar returns. 44/ Special Investigation Report--"Runway Ineursions at Controlled Airports in the United States" (NTSB/SIR~-86/01). 15/ Schroeder, D. --- Footnotes: [15/ Schroeder, D. J., "The loss of prescribed separation between aircraft: How does i occur?" Transcripts, 1985 Conference of the Soclety of Automotive Engineers, 4426-4434,] [44/ Special Investigation Report--"Runway Ineursions at Controlled Airports in the United States" (NTSB/SIR~-86/01).]

[ANALYSIS] The Safety Board believes that the LDB symbology assoclated with the radar target of N60SE was sufficiently prominent on the controllers' radarscopes that the controllers should have seen it. However, reliance on ARTS FDB radar symbology may have been responsible for their failure to see the target symbology associated with N60SE. If this type of oversight Is occurring elsewhere in the ATC system, controllers are denying themselves radar target information that would potentially reduee the continuing threat of midair collisions between IFR and VFR aircraft. The Safety Board believes that the FAA should examine the underlying ATC factors in midair collisions and near-midair collisions to determine the extent to which controllers have become dependent on ARTS FDB symbology and the training or remedial measures needed to alleviate the problem. The failure of the East Radar controllers to provide timely traffic advisories and a safety alert to the crew of the U~21 placed that IFR flight al the same midair collision risk as VFR aircraft which were not using FAA flight~: sllowing services. The Army pilots, perhaps unknowingly, became completely dependent on their own ability to "gee and avoid" other airplanes, with all the inherent limitations of the "see aad avoid" method of avoiding inflight collisions. At the same time, they had reason to expect that the radar controllers were not particularly busy (not much radio cammunication and excellent weather conditions) and would alert them if there was conflicting traffic. Sueh reasoning would have been reinforced by the traffic advisory provided to them about 3 minutes before the collision. Under the circumstances, the pilots may have been particularly vuinerable to such an accident because of their confidence in the ATC systen. Unfertunately, ATC systems in the United States are not equipped with an automated system that would alert the radar controtlers to the presence of a conflict between an iFR flight and a mode-C transponder~equipped VFR flight. The East Radar controllers were unable to provide the needed traffic advisory information because they did not detect the threat, even though the information they needed was displayed before them, and because their ARTS computer was not equipped with the programming that wuld have alerted them to the threat. Although the Safety Board cannot state with certainty that the pilots would have taken timely and appropriate action 12 avert the accident if they had received trafic advisory information, the Safety Board v.'ieves that the Army pilots’ chanees of averting the collision would have been improved substantially if such information had been provided. Any information the controllers could have provided to the U-21 pilots would have improved the crew's probability of acquisition of the PA~31 over that of an unalerted flightcrew.

[PROBABLE CAUSE] Aviation Administration: Update Advisory Circular 90-48C and emphasize in operational bulletins, the Airman's Information Manual, pilot training programs, and aceident prevention programs the advantages of using air traffic contro! flight~following services on visual flight rules flights as a further means of reducing the midair collision hazard. (Class JI, Priority Action) (A-88-24) Incorporate formal training on the dangers of the low-workload environment ut all levels of alr traffic controller training. (Class fl, Priority Action) (A-88-25) Establish an ad hoc task foree, including controller and human performance expertise, to evaluate the extent to which radar air traffic ecxtroliers are dependent on FDB radar synibology to carry out thelr duties and to make appropriate improvements in initial and cecurrent radar training to rectify such deficiencies. (Class Il, Priority Action) (A-88- 26) Expedite the development, certification, and production of various low-cost proximity warning and conflict detection systems for use aboard general aviation alreraft. (Class I, Priority Action) (A-88-27) --to the National Business Aircraft Association and the Aireraft Owners and Pilots Association: Make the facts and circumstances of this acelident known to your membership and encourage the use cf the services of the air traffic control system as a means 0; reducing the potential for midair collisions. (Class H, Priority Action) (A-88-28) BY THE NATIONAL TRANSPORTATION SAFETY BOARD /sf SIM BURNETT Chairman /sf PATRICIA A, GOLDMAN Vice Chairman /s/_ JOHN K, LAURER Member /sf JOSEPH 2. NA! LL Member JAMES L. KOLSTAD Member February 3, 1988 5, APPENDIXES APPENDIX A INVESTIGATION AND HEAZING 1. (investigation The Safety Board's Kansas City Field Office was initially notified of the accident ubout 1300 central standard time, January 20, 1987 and immediately responded to the accident seene before it was known that two aircraft were involved, About 1700 eastern standard time, the Safety Board was notified by the Federal Aviation Administration that a second aireraft was involved and that there had been a midair collision. Early on January 21, 1987, three additional investigators were dispatched to the scene from Washington, D.C., to participate in the accident investigation. Parties to the investigation were the Federal Aviation Administration, Sachs Electric Company, the U.S.

[ANALYSIS] AC 90-48C urges VFR pilots to take advantage of air traffic advisory services as a means of assisting them in seeing and avoiding other aircraft, but not substituting for the pilots' own visual scanning. The AC was tssued before the conflict alert feature was in widespread use in the U.S. ATC system. Although the Safety Board eoneurs with the emphasis that the AC places on pilots scanning effectively to avoid midalr collisions, the Safety Board belleves that AC 96-48C should be updated to alert pilots to the significant additional safety benefits accruing from conflict alert when flight~following services are provided to VFR pilots. In addition to the factors already discussed, the Safety Board considered other factors that: could have influenced the pilots’ ability to effectively scan the sky for potential midair collision threats, Those factors included conspiculty of the target, task variables, distractions including occupation with other crew duties, visibility restrictions due to environmental conditions (including snow cover on the ground) and the condition of the windshield glass, pilot fatigue, and empty field myopia (a tendency for the human eye to focus at arms length until objects are identified at a greater distance) It was considered probable that decreased vigilanee on the part of the U-21 flighterew, who had been at the controls for nearly 3 hours, and occupation with normal cockpit duties on the part of the pilots of both airplanes may have reduced the degree of outside seanning that was occurring as the airplanes converged. The condition of the windshield glass was not known. Any of these factors would have reduced the time in which the pllots were actively or effectively attempting to “see and avoid." Since {{It is uncertain at :xactly what point the pilots could have visually acquired the other airplane, the Safety Board is unable to state with certainty that the pilots could not have avoided the collision. However, the Safety Board believes that this ~2R- case demonstrates limitations cf the "see and avoid" concept that would have impeded significantly the pilots! collective ability to avoid the collision. The Safety Board believes that the see and avoid concept alone may not have been sufficient to avert this accident and that an additional safeguard in the form of automated ATC system redundancy is needed to prevent such midair collision accidents. 2.5 ATC Services The procedures contained in the controller's Handbook require controllers to set priorities on the services they provide to aireraft with first priovity given to the separation of [FR airplanes and the provision of safety alerts. The Handbook states that traffle advisories, distinguished from safety alerts by their lesser urgency, are provided as an additional service, "workload permitting" and "contingent upon higher priority duties." In this case, workload should not have impalved the ability of the controllers to provide additional services; the traffic was light and the operational situation was not complex. Even so, the controllers at the East Radar position reported that they never observed any target in the vicinity of Army 18)61 in the minutes before the collision.

[PROBABLE CAUSE] The Basic Flight Information and ATC Procedures describe the airspace segments and the basic pilot responsibilities for operating in such airspace, (vi) Contact the nearest FAA Flight Service Station for any pertinent . NOTAMS pertaining to their area of operation. EE SMe Uran On estrone eens Wr PA Ts aes a ia wb PB RI SERS ‘ daweed ophratn 8 ¢E MRS EONS TSE BP NS ag eee 9 es os APPENDIX F AC 90+48C 5118/83 (3) Pilots should also be familiar with, and exercise caution, in those operational environments where they may expect to find a high wiume of traffic or special types: of aircraft operation, These areas include Terminal Radar Service Areas (TRSA'sS), airport traffic patterns, particularly at airports without a control tower; airport traffic areas (below 3,000 feet above the surface within five statute miles of an airport with an operating contro] tower); terminal control areas; control zones, including any extensions; Federal airways; vicinity cf VOR's; restricted areas; warning areas; alert areas; Military Operating Areas (MOA); intensive student jet training areas; military low-level high-speed training routes; instrument approach areas; and areas of high density jet arrival /depa ‘ture routings, especially in the vicinity of major terminals and military bases. e, Use of Communications Equipment and Air Traffic Advisory Services. (1) One of the major factors contributing to the likelihood of NMAC incidents in terminal areas that have an operating air traffic control (AIC) system has been the mix of known arriving aml departing aircraft with unknown traffic. The known aircraft are generally in radio contact with the controlling facility (local, approach, or departure control) and the other aircraft are neither in two-way radio contact ror identified by ATC at the time of the NMAC, This precludes ATC from issuing traffic advisory information to either aircraft. (2) Although pilots should adhere to the necessary communications requirements when operating VFR, they are also urged to take advantage of the air tratfie advisory services available to VFR aircraft. {{3}} Psiots should: (i) Use the AIM. {{4}} The basic AIM contains a section dealing with services available to pitots, including information on VFR advisory services, radar traffic information services for VFR pilots, and recommendee traffic advisory practices at nontower airports, (B) The airport/facility directory contains a iist of all major alyports showing the services available to pilots and the applicable communication frequencies. (ii) Develop a working knowledge of those facilities providing traffic advisory services and the area in which they give these services. (1ii) Initiate radio contact with the appropriate terminal radar or nonradar facility when operating within the perimeters of the advertised service areas or within 15 miles of the facility wien m service area is specified. (iv) When it 16 not practical to initiate radio contact for traffic information, at least monitor the appropriate facility communication frequency, particularly when operating in or through arrival/departure routes and instrumer.t approach areas, APPENDIX F 3/18/83 AC 90-48C (vy Remember that controller cbeervat.ion of aircraft in the terminal area is often limited by distance, depth perception, aircraft conspicuity, and other normal visual acuity problems, Limitations of radar (when available), traffic volume, controller workload, unknown traffic, etc., may prevent the controller from providing timely traffic advisory information.

[ANALYSIS] Even so, the controllers at the East Radar position reported that they never observed any target in the vicinity of Army 18)61 in the minutes before the collision. Obviously, if the information related to N6Q0SE did not appear or was not perceived on their radarseope, traffic advisories or a safety alert would not have been provided to the pilot of the Army airplane. The Kansas City TRACON was not in communieation with N6OSE; thus, there was no opportunity to provide traffic advisories to that airplane. The East Radar controllers reported that if they had seen the radar target of an aircraft that represented a threat to Army 18061, they would have provided the appropriate traffic advisories. However, review of the recorded radar data and TRACON communications revealed that in the 7 minutes before the accident, traffic advisories were only provided to aireralt presenting an FDB on the controllers' scope; such advisories were only provided regarding traffic represented on the radar screen by an FDB, ‘The Safety Board was unable to establish that the eontrollers would have overlooked traffic represented by an LDB or that they wouJd have unintentionally given a lower priority to traffic represented by an LDB. Howe:.er, the recorded radar data and data from the retrack program suggesred that the radar information relevant to N60SE was recorded, processed, and presented on the controllers’ scope. The maintenance records, a postaccident ground check, and a flight cheeix of the East Radar equipment and radarscope did n-c reveal any indication of a discrepancy that would have prevented the presentation of the LDB of N6GSE on the East Radar controller scope at the time of the accident. Therefore, the Safety Board concludes that the radar target of N60SE was displayed on the East Radar controller scope; yet both controllers failed to perceive it and the collision threat represented by it in the minutes before the accident. This failure elevates the concerns of the Safety Board that ATC system redundancy in the form of VFR conflict alert programming is needed to assist in the prevention of such midair collision accidents. As a result of this and three other midair collision accidents, on July 27, 1987, the Safety Board recommended that the Federal Aviation Administration: Take expedited action to add visual flight rules conflict alert (mode C intruder) logie to Automated Radar Terminal System (ARTS) IU) A systems as an interim measure to the ultimate implementation of the Advanced Automation System. ee ee erent eee ene i RAE TOTLSS o fhe timing and completeness of the position rellef briefing given 1o the developmental controller by her area supervisor shortly before the accident nay have been of critical importance in this accident.

[PROBABLE CAUSE] Traffic advisories are secondary to the controllers' primary duties (which are separating aircraft under their control and issuing safety advisories when aware of safety conflicts). Therefore, the pilot is responsible for seeing and avoiding other traffic. Traffic advisories should be requested and used when available to assist the pilot to see and avoid other traffic by assisting, but not substituting in any way, the pilot's own visual scanning. It is important to remember that advisories which air traffic control may provide are not intended to lessen in any manner the pilot's obligation to properly scan to see and avoid traffic. f. Airport Traffic Patterns. (1) A significant number of midair collisions, as well as near midair collisions, have. occurred within the traffic pattern envirerment. (2) Pilots should: (i) When operating at tower-controlled airports, maintain two-way radio contact with the tower while within the airport traffic area. Make every effort to see and properly avoid any aircraft pointed out by the ‘cower, or any other aircraft which may be in the area and unknown to the tower. (11) When entering a known traffic pattern at a nontower airport, keep a sharp lookout for other aircraft in the pattern. Enter the pattern in level flight and allow plenty of spacing to avoid oertaking or cutting any aircraft out of the pattern. (iii) When approaching an unfamiliar airport fly over or circle the airport at least 500 feet above traffic pattern altitude (usually at 2,000 feet or more above the surface) to observe the airport layout, any local traffic in the area, and the wind ané@ traffie direction indicators. Never descend into the traffic pattern from diractly above the airport. (iv) Be particularly alert before turning to the base leg, final approach course, and during the final approach to landing. At nontower airports, avoid entering the traffic pattern on the base leg or from a straight-in approach to the landing runway. {{v) Canpensate for blind spots due to aircraft design and flight attitude by moving your head or maneuvering the aircraft. g.

[CONCLUSIONS > FINDINGS] The pilots of both airplanes were qualified and were familiar with the Kansas City area. There were no apparent medical factors influencing their performance. Both airplanes were airworthy. There were no apparent airplane equipment deficiencies or system malfunctions. "‘1e accident occurred in visual meteorological conditlons where the pilots of both airplanes were required to “see and avoid" the other. There was no Indication that elther pilot took evasive action to avoid the collision. Both airplanes were equipped with operating mode-C transponders. The U-21 was operating under [FR and the PA-31 was operating under VFR. The U-21 was displayed as an FDB and was a computer-tracked target on the Kansas City International TRACON controllers' radarscope. The PA~31 was displayed as a code 1200 LDB with mode-C altitude information on the same controllers! radarscope. Although the East Radar position was staffed by two controllers, neither observed any target in the vicinity of the data block representing the U-21, The conflict alert subprogram of the ARTS lll tracking system was not programmed te alert them of an impending collision invalving an IFR aircraft and an untracked VFR aircraft. / Special Investigation Report~-"Midair Collisions In U.S. Civil Aviation, 1969-1970" TSB/A AS-72/6). 4 Three minutes before the accident, the U-21 was provided a traffie advisory concerning another airplane. Traffic advisories concerning the PA-31 were not provided. 11. The area supervisor had just briefed and was providing instruction to a developmental controller at the East Radar position when the collision occurred, The controller workload at the East Radar position was light. The PA-31 pilot did not use VFR flight-following services that were available to him. Conflict alert would have alerted the East Radar controllers to the collision threat involving the airplanes 40 seconds before the collision if the PA-31 had been a tracked target. 13. The "see and avoid" concept provided marginal opportunity to the pilots of both airplanes to avert the collision. 14, The absence of VFR conflict alert logic in ARTS Ill equipment at Kansas City diminished the potential for the radar controllers to detect the impending conflict. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of the accident was the failure of the radar controllers to detect the conflict and to issue traffic advisories or a safety alert to the flightcrew of the U-21; deficiencies of the see and avold concept as a primary means of collision avoidance; and the lack of automated redundanvy in the alr traffic control system to provide conflict detection between participating and nonparticipating aircraft. 4, RECOMMENDATIONS As a result of this and three other midair collision accidents, on July 27, 1987, the National Transportation Safety Board recommended that the Federal Aviation Administration: A-87-98 Take expedited action to add visual flight rules conflict alert (mode C intruder}} logie to Automated Radar Terminal System (ARTS) Ill A systems as an interim measure to the uitimate implementation of the Advanced Automation System, As a result of its investigation of tnis accident, the National Transportation Safety Board recommended: ~-to the Federa!

[ANALYSIS] The Safety 1i/ Billings, C., Grayson, R., Heeht, W., and Curry, R., "A Study of Mear Midair >ollisions in U.S. Terminal Al: ~ace," NASA Technical Memorandum 81225, 1980. 12/ Duffy, E., "The Psycholc,,.cal Significance of the Concept of Arousal or Activation," Psychological Review, 1957. re : BAS LISTE IEE Myles Spars Or oN TERR ER SAO tS REA ASE Be WAS 31 RY ORE EAL SEE CLE A OR OD SR RRR re Board believes that her qualification on two other radar positions and training reecived on the East Radar position attests adequately to her knowledge of ATC radar preeedures, ever though she ‘vas not yet qualified on the Kast Radar position. It was not inapprepriate for a deveiopmenta! controller to be reeciving OJT from her imimediate supervisor. The area supervisor who briefed the developmental controller on the position and then monitored her performance was appropriately experienced and qualified to perform the OJT function. The Safety Board is aware that the area supervisor was obligated to monitor the training progress of the devclopmental controller and, based on satisfactory performance, to eventually certify her on the position. The Safety Board is concerned, nonetheless, that the area supervisor, because of his preoccupation with the briefing of the developmental controler, may not have given appropriate attention to the East Radar position in the moments before the accident. The accident o¢eurred outside the bcundaries of the Kansas City TCA, Therefore, the applicabie rules and the safety benefits associated with that protected airspace are not relevant to this accident. The collision oceurred in airspace where ATC separation services were provided to IFR aircraft and traffic advisories were provided to VFR aircraft receiving flight-following services. Except for the fact that under Federal aviation rules, VFR aireraft were obligated to remain in visual meteorological conditions and to see and avoid other aireraft operating in tnat same airspace, VFR aircraft were authorized to operate in the airspace outside of a TCA without receiving ATC se-vices. Since visual meteorological conditions were prevalent, it was not inappropriate for both airplanes io have been operating in the airspace where the collision occurred. Thus, the Safety Board's analysis first examined the collision geometiy to evaluate the potential for the pilots to see and avoid each other. The collision geometry was reconstructed from the physical evidence found in the wreckage of the airplanes and from ARTCC and ARTS Ili radar data. --- Footnotes: [12/ Duffy, E., "The Psycholc,,.cal Significance of the Concept of Arousal or Activation," Psychological Review, 1957.]

[ANALYSIS] Any information the controllers could have provided to the U-21 pilots would have improved the crew's probability of acquisition of the PA~31 over that of an unalerted flightcrew. The Safety Board believes that the failure of the controlers to perceive the collision threat and to provide traffic advisory information was so important to avoidance of the collision that it is cited as a cause of ‘his accident along with limitations in the ATC system that made it difficult for the controllers to distinguish collision threats between JFR and VFR alreraft. 2.6 Prevention of the Avcident The retrack of tne Kansas City TRACON ARTSIII data demonstrated graphically how this accident might have been prevented. Ry manually tagging up the LDB of the PA~31 (during the retrack), an FDB was generated and computer tracking of the PA~31 was initiated automatically. This activated the conflict alert subprogram of the ARTS IH equipment. The conflict alert subprogram compared the progress of the flight track and altitude information of the PA-31 with that of all other tracked targets. Then about 40 seconds before the collision, an aural alarm was activated; the data block information of the conflicting targets began to flash on the controller's radavscope; and a conflict alert message identifying the airplanes in conflict was displayed in the preview area of the radarscope. The Safety Board belleves that If this type of distinet and unambiguous information had been presented to alert the controllers before the accident, the controller's attention would have been immediately focused on the conflicting airplanes, and the controlier would have had ample opportunity to issue a traffic advisory or a safety alert to the U-21 pilots. The Kansas City TRACON East Radar controllers did not have the benefit of conflict alert before the accident because VER airplanes are not provided diserete transponder codes, are not tagged up (tracked) unless they request and are provided ATC flight-following services, and because con‘liet alert programming does not provide a warning to contrellers when a conflict belween an IFR asireraft and an untracked VFR code 1200 target occurs. Transponder-equipped aircraft on VIR flights normally broadeast code 1200 to inform controllers of their location and VER status. An LDB is then presented on the controller's radarseope if the transponder has made C. The retrack cemonstrated that if flight-following services had been provided to the PA~31 pilot, the conflict alert subprogram would have alerted the controllers to the colliston threat about 40 seconds before the collision, (Obviously, if the application of the conflict alert subprogram could be extended to include VFR mode-C aireraft, air traffic controllers could extend more pcsitive protection against the threat of midair collisions and to a much larger population of aircraft than are protected by the present conflict alert system.

[ANALYSIS] J., "The loss of prescribed separation between aircraft: How does i occur?" Transcripts, 1985 Conference of the Soclety of Automotive Engineers, 4426-4434, rg nnd toa i ST aa oe BBE ar a i ek ti Ro i a C2 ema fs ens rT In conjunction with this investigation and the Safety Board's investigation of the January 15, 1987, Kearns, Utah, midair collision accident, the Radar Training Facility at the FAA Academy in Oklahoma City, was examined. The Safety Board attempted to determine whether there were deficiencies in the training of air traffic controllers that would explain their not detecting collision threats represented on radarseopes by primary, secondary, or LDB rader targets, as opposed to FDB radar targets. A review of the radar controller curriculum and laboratory exercises at the FAA Academy revealed that they were sufficient in terms of exposing controllers to the radar situations described in their Handbooks. The laboratory workshops allow controllers to gain practical knowledge and experience in the application of radar procedures, including the provision of traffic advirories. Controllers were graded on their demonstrated ability to perceive and react to VFR traffic and in making appropriate traffic advisories. However, it was noted in this program that emphasis was placed on the appropriate separation of FDB IFR traffic, and when VFR traffic was Introduced, it was always represented by LDB radar targets with altitude information displayed. Thus, it was not possible for developmental radar controllers at the FAA Academy to detect conflicts involving VFR targets represented by primary or secondary radar targets only. ARTS tracking systems superimpose computer~generated alphanumeric symbology over the primary and secondary radar target information on controller radarscopes. Tracking of radar targets and distinguishing IFR from VFR targets is much easier using the ARTS information than the primary and secondary radar information that is also displayed. Because FDBs provide more alphanumeric symbology (and inforination) than LDBs, and because radar controllers control traffic that is almost always identified by FDOBs, there is reason to believe that LDBs might soinetimes be overlooked by controllers, particularly when controller workload js high or when controller vigilance is reduced. Controller dependence on ARTS IJJI FDB target symbology could cause controllers to attach diminished importance to primary, secondary radar, and LDB target information even when mode~C transponder information is provided. The Safety Board believes that the LDB symbology assoclated with the radar target of N60SE was sufficiently prominent on the controllers' radarscopes that the controllers should have seen it. However, reliance on ARTS FDB radar symbology may have been responsible for their failure to see the target symbology associated with N60SE.

[ANALYSIS] Pilots could also avail themselves of the benefits of the confliet alert subprogram if they would use flight-following services on VFR flights when those services are available. Although controllers are not always able to provide flight~following services to VFR aircraft because of workload, there is no reason to believe that such services would not have been provided to N60SE, since traffie was light. Interviews of the Kansas City TRACON staff and review of their policies indicated that air traffic services typically would have been provided to the PA-31 pilot under the circumstances of the accident flight had those services been requested. During heavy controller workload conditions and at facilities which are normally very busy, VFR pilots may find that their requests for flight-following or other ATC services are frequently not fulfilled. Recognizing that the workload of many faciilties is already high at times and would be increased to the extent that some VFR pilots may not always be able to obtain alr traffic services, the Safety Board believes that VFR pilots should nonetheless attempt to obtain those services, when they are available, as a means of reducing the potential for involvement in midair collisions. In this case, the Safety Board concludes that the accident probably would have been prevented if the PA-31 pilot had availed himself of flight-following services (or filed an IFR flight plan). While acknowledging that the excellent record of midair collision prevention, particularly within positive control airspace and in TCAs, is a tribute to controller performance, the Safety Board believes that controllers need additional automated redundancy to assist them in their task, Additionally, pilots need a more positive backup to the "see and avoid" concept of collision avoidance. Conflict alert and improvements in terminal facility computer systems have provided automated assistanec, but do not presently allow controllers to identify collision threats which involve untracked VFR aircraft. Many ATC operational errors and serious compromises of separation between IFR aircraft have been prevented because of the conflict alert feaiu.e. The Safety Board has been advised by the FAA that a TRACON ARTS IIIA computer could be expanded by adding processing capability to include VFR mode-C intruder conflict alert logic. (The FAA plans to upgrade all ARTS IIl terminal facilities, including the Kansas City TRACON, to the ARTS ill A capability.) The Safety Board recognizes that the procurement of additional processors could infringe on other FAA priorities and may be viewed as an interim measure to the future installation of the Advanced Automation System which is due to be Implemented in the late 1990s.

[ANALYSIS] The possible effect of the high winds on the indicated altitude has been discussed previously. While the evidence does not indicate whether the crew was aware of the SIGMETS issued for the Washington area, there is no evidence to indicate that knowledge of the SIGMETS would have caused the crew to operate any differently than they did. "= 35 - The CVR indicates that the crew did encounter considerable turbulence during the descent. However, the record also indicates that they.were able to read the altimeters well enough to know that they had descended below their target altitude of 1,800 feet. The Safety Board believes that the effect of turbulence was not critical but could not determine positively why the descent was not arrested at 1,800 feet. . In summary, this accident resulted from a combination of . conditions which included a lack of understanding between the controller and the pilot as to which air traffic control criteria were being applied to the flight while it was operating in instrument meteorological conditions in the terminal area. Neither the pilot nor the controller understood what the other was thinking or planning when the approach clearance was issued. The captain did not react correctly to his own doubt about the line of action he had selected because he did not contact the controller for clarification. The action of the other air carrier pilot who questioned the clearance he received about 1/2 hour before the accident is the kind of reaction that should be expected of a pilot suddenly confronted with uncertainty about the altitude at which he should operate his aircraft. The Board again stresses that it is incumbent upon air carrier management to assure the highest possible degree of safety through an assertive exercise of its operational control responsibility. This management function must assure that flightcrews are provided with all information essential to the safe conduct of flight operations. Furthermore, the air carrier must assure that its flightcrews are indoctrinated in the operational control precept and that during flight the final and absolute responsibility for the safe conduct of the flight rests solely with the captain as pilot-in-command regardless of mitigating influences which may appear to dilute or derogate this authority. Whereas the air carriers and the pilots are expected to per-— form their services with the highest degree of care and safety, this same high level of performance must be expected from the management of the air traffic control system and the controller. The instant case provides a classic and tragic example of a pilot and controller who did not fully comprehend the seriousness of the issuance and’ acceptance of a clearance which was not precise or definitive. The pilot should question a clearance which leaves any doubt as to what - 36 -. 13/ course of action should be followed. —- The Board also believes that it is incumbent upon the controller to ascertain beyond a doubt that the terminology of a clearance conveys the intent to the pilot, and to question the pilot if there is any doubt that he has understood it and is initiating actions compatible with the intent of the clearance. --- Footnotes: [13/ course of action should be followed. —- The Board also believes that it is incumbent upon the controller to ascertain beyond a doubt that the terminology of a clearance conveys the intent to the pilot,]

[CONCLUSIONS] During the course of the investigation it became clear that there was an omission in the ATC handbook concerning exactly when radar service is terminated. It was unfortunate that the handbook did not clearly ~ 44 - require the controller to provide altitude restrictions when an aircraft is operating over an unpublished route for which there is no minimum enroute altitude prescribed, while the flight was being handled as a non-radar arrival. The FAA has since limited such clearances and some action is being taken to correct the deficiencies cited above. “Yond A. Ss, Isabel A. Burges Member December 2, 1975 -~ 45 - McADAMS and HALEY, Members, dissenting: We do not agree with the probable cause as stated by the majority. In our opinion, the probable cause was the failure of the controller to issue altitude restrictions in accordance with the Terminal Air Traffic Control Handbook 7110. 8C, paragraph 1360(c), and the failure of the pilot to adhere to the minimum sector altitude as depicted on the approach plate or to request clarification of the clearance. Asa result, the pilot prematurely descended to 1, 800 feet, The flight was a radar arrival and, the refore, entitled to altitude protection and terrain clearance. If the controller, as required by the thenexisting procedures for radar arrivals, had issued altitude restrictions with the approach clearance or had deferred the clearance, the accident probably would not have occurred, On the other hand, if the pilot had either maintained the minimum sector altitude of 3, 300 feet as depicted on the approach plate, or requested clarification of the clearance, there would not have been an accident, ‘ The majority states (p. 32): "The Board concludes that based on the criteria in 7110. 8C the system allowed for the classification and handling of Flight 514 as a nonradar arrival. The Board, however, believes that the flight should have been classified and handled as a 'radar arrival,'"' This statement cannot be reconciled with the probable cause as stated by the majority. If the majority believes that under all tLe circum - stances the flight should have been classified and handled as a radar arrival, then the flight was in fact a radar arrival and the probable cause should so state. Itis not possible to determine from the majority opinion whether Flight 514 was a tadar or a nonradar arrival. The Board attributes the failure of the controller to handle the flight asa radar arrival to be a terminology difficulty between pilots and controllers. There was no terminology difficulty. .The plain fact of the matter is that the controller simply did not treat the flight as a radar arrival as he should have, All the criteria of paragraph 1360 for a radar arrival were present. Neither the pilot nor the controller had terminology difficulties. The pilot assumed he was a radar arrival and would be given altitude restrictions ifnecessary.

[CONCLUSIONS] There was no terminology difficulty. .The plain fact of the matter is that the controller simply did not treat the flight as a radar arrival as he should have, All the criteria of paragraph 1360 for a radar arrival were present. Neither the pilot nor the controller had terminology difficulties. The pilot assumed he was a radar arrival and would be given altitude restrictions ifnecessary. Not having received such restrictions, he initiated a descent to 1, 800 feet. ~ 46 - Additionally, the Board concludes on the subject of radar arrival (p. 32): ",,.under these circumstances, [the clearance] should have included an altitude restriction until the aircraft had reached a segment of the published approach procedure or the issuance of the approach clearance should have been deferred until the flight reached such segment. . Therefore, the Safety Board concludes that the clearance was inadequate and its issuance and acceptance was the result of a misunderstanding between the pilot and the controller." Such a conclusion can again only mean that the flight was in fact a radar arrival since altitude restrictions are issued only in accordance with paragraph 1360(c), the provisions of which pertain solely to radar arrivals. Therefore, based upon the foregoing, it would appear the majority believes the flight was a radar arrival but refuses to make an unambiguous finding to that effect. . The Board further states (p. 32) that ''there is a general lack of understanding between pilots and controllers in their interpretations of air traffic control procedures.'' We find that there was no misunderstanding in this instance on the part of the pilot. As previously stated, he undoubtedly descended to 1, 800 feet after receiving an approach clearance because he was not issued an altitude restriction. If the controller was confused with regard to the application of paragraph 1360,he should have _asked for clarification from his supervisor. But there should have been no reason for confusion insofar as terminology is concerned. One of the most important functions of an air traffic controller is to possess the highest degree of knowledge in procedures and terminology and to apply it with the greatest diligence and care. . In any event, we can only conclude that, innot handling the flight as a radar arrival, the Dulles controller did not properly apply the provisions of the controller's handbook.

[ANALYSIS] Pilots should not be faced with the necessity of choosing from among several courses of action to comply with a clearance. . The Board believes that the clearance, under these circumstances, should have included an altitude restriction until the aircraft had reached a segment of the published approach procedure or the issuance of the approach clearance should have been deferred until the flight reached such segment. Therefore,the Safety Board concludes that the clearance was inadequate and its issuance and acceptance was the result of a misunderstanding between the pilot and the controller. $ The Board believes that there is a general lack of understanding between pilots and controllers in their interpretations of air traffic control-procedures. There is also a lack of understanding about the meaning of some words and phrases used by both the controller and pilot in the handling of IFR traffic in the terminal area. In this case, there was no definition of the term "radar arrival" or "final approach course," nor, as indicated earlier, did there seem to be common understanding between pilots and controllers as to the meaning of "radar control. "' Therefore, the Safety Board concludes that it is essential that a lexicon of air traffic control words and phrases be developed and made available to all controllers and pilots who operate within the National Airspace System. Additionally, there should be one book of procedures for use by both pilots and controllers so that each will understand what to expect of the other in all air traffic control situations. This manual must be used in the training of all pilots and controllers. The need for such a lexicon and procedures manual is evident from the circumstances of this accident. Flight 514 was vectored to intercept the 300° radial of Armel, the reciprocal course of which coincides with the course for the intermediate and final approach segments of the published instrument approach procedure. The vector was given when the flight was more than 80 miles from the airport and at a point where the 300° radial of Armel was not a part of the published instrument approach procedure. While proceeding inbound on the 300° radial ui Armel, the flight would not have reached a segment of the published approach procedure until it arrived at Round Hill. However, there was some testimony contending that Flight 514 was on its final approach course when the flight intercepted and was inbound on the 300° radial, and accordingly it was permissible for the pilot to descend to the minimum altitude of 1,800 feet prescribed for crossing the final approach fix of the VOR/DME instrument approach procedure. Qualified instrument pilots and air traffic controllers should know and understand beyond equivocation that the coincidence of the inbound course being an extension of the final instrument approach course does not permit ‘descent to altitudes lower than those published for that air space segment unless specifically authorized by ATC. A clear, precise definition of final approach course and final instrument approach course should preclude future misunderstandings. Neither of these terms was defined in the AIM at the time of this accident.

[CONCLUSIONS] Another crew that questioned a similar clearance received further instructions and information which resulted in their accepting a radar surveillance approach to Dulles. ) Both military and civil aviation officials for several years had indicated concern regarding a lack of understanding on their part of what the Air Traffic Control procedures and terminology were intended to convey to the pilots. They were also concerned about the possibility of misunderstandings which could result in pilots descending prematurely. ' The FAA was not responsive to the long standing, expressed needs and concerns of the users of the Air Traffic Control System with regard to pilot/ controller responsibilities pursuant to the issuance of an approach clearance for a nonprecision approach. Furthermore, the FAA did not provide users of the Air Traffic Control System with sufficient information regarding the services provided by the system under specific conditions. The FAA did not utilize the capability of the ARTS III system to insure terrain clearance for descending aircraft conducting nonprecision instrument approaches in instrument meteorological conditions. The flightcrew of Flight 514 was not familiar with the terrain west and northwest of Dulles. However, they did have information regarding the elevation of obstacles west of Round Hill intersection depicted on the plan view of the approach procedure. b.. Probable Cause The National Transportation Safety Board determines that the probable cause of the accident was the crew's decision to descend to 1, 800 feet before the aircraft had reached the approach segment where that minimum altitude applied. The crew's decision to descend was a result of inadequacies and lack of clarity in the air traffic control procedures which led to a misunderstanding on the part of the pilots and of the controllers regarding each other's responsibilities during operations in terminal areas under instrument meteorological conditions. Nevertheless, the examination of the plan view of the approach chart should have disclosed to the captain that a minimum altitude of 1, 800 feet was not a safe altitude. ‘ , Contributing factors were: (1) The failure of the FAA to take timely action to resolve the confusion and misinterpretation of air traffic terminology although the Agency had been aware of the problem for several years; (2) The issuance of the approach clearance when the flight was 44 miles from the airport on an unpublished route without clearly defined minimum altitudes; and (3) Inadequate depiction of altitude restrictions on the profile view of the approach chart for the VOR/DME approach to runway 12 at Dulles International Airport. 3, RECOMMENDATIONS As a result of the accident, the Safety Board submitted 14 recommendations to the Administrator of the Federal Aviation Administration. (See Appendix I.) Subsequent to the accident, the FAA has taken several actions in an effort to prevent recurrence of this type of accident. 1. The FAA has directed that all air carrier aircraft be equipped with a ground proximity warning system by December 1975. 2. The FAA has revised the provisions of 14 CFR 91 with regard to pilot responsibilities and actions after receiving a clearance for a nonprecision approach. 3. The FAA has established an incident reporting system which is intended to identify unsafe operating conditions in order that they can be corrected before an accident occurs. .~ 4] - 4. The FAA has changed its air traffic control procedures to provide for the issuance of altitude restrictions during nonprecision instrument approaches. 5. The FAA is installing a modification to the ARTS III system that will alert air traffic controllers when aircraft deviate from predetermined altitudes while operating in the terminal area.

[ANALYSIS] Particular emphasis was made by FAA that the vector to the 300° radial occurred when the flight was approximately 80 miles from the airport and that it was vectored by the center on to an en route course. Operational advan_ tage was indicated by the controllers as the reason for the vector to the 300° radial rather than to an initial approach fix on the approach procedure. , The counterposition is that Flight 514 was operating ina .radar environment, was receiving at least one type of radar service, and was on a course which would lead directly to the Round Hill intermediate approach fix. Furthermore it had been advised that the reason for the vector to the 300° radial was for a VOR/DME approach for runway 12. Consequently, it should have received services, including altitude restrictions, as set forth in Paragraph 1360 of 7110. 8C. , In evaluating these facts, the one issue present is whether the handling of Flight 514 required the provision of an altitude restriction. FAA witnesses agreed that, had Flight 514 been classified as a radar arrival within the meaning of the handbook, the flight would have been given an altitude restriction until it reached Round Hill. In resolving this issue, the Board has been troubled by the fact that ATC procedures are almost always dependent upon the usage of certain specified phrases and terms, many of which have no established definitions and mean different things to controllers and pilots. The term "radar control''is an example. The pilot witnesses believed that, when they were operating in a traffic control radar environment, they were being controlled by radar. The controller group was aware that this was not always the case, but the FAA apparently did not perceive the difference of understanding, and the efforts made by the FAA to clarify when an aircraft was or was not radar controlled did not eliminate the confusion. The Board concludes that based on the criteria in 7110. 8C the system allowed for the classification and handling of Flight 514 as a nonradar arrival. The Board, however, believes that the flight should have been classified and handled as a "radar arrival." This, however, does not dispose of the issue of whether the ATC system should have provided for a redundancy that would have prevented or consequently identified and corrected a deviation of an aircraft from a clearance which was not followed as the controller expected it to be. The system should clearly require controllers to give the pilots specific information regarding their positions relative to the approach fix and a minimum altitude to which the flight could descend before arriving at that fix. Pilots should not be faced with the necessity of choosing from among several courses of action to comply with a clearance. . The Board believes that the clearance, under these circumstances, should have included an altitude restriction until the aircraft had reached a segment of the published approach procedure or the issuance of the approach clearance should have been deferred until the flight reached such segment.

[CONCLUSIONS] The first officer's altimeter was set properly. It is possible that wind velocity over the hilly terrain may have induced an altimeter error which could have caused the instrument to indicate that the aircraft was higher than its actual altitude. However, the crew's last.comments regarding altitude indicated that they knew they were below 1,800 feet. The altitude alerting system and the radio altimeter aural warnings sounded at appropriate altitudes to indicate to the pilots that the aircraft was below 1, 800 feet and that the aircraft was within 500 feet and 100 feet of the ground. These latter warnings occurred 7 seconds and 1 second, respectively, before impact. The flightcrew apparently did not have sufficient time to avoid the accident after these warnings. 13. 14, 15... 16. 17. 18. 19. The approach clearance was given to the flight without altitude restrictions because the flight was not being handled as a radar arrival and because the controller expected the crew to conduct the approach as it was depicted on the approach chart. Procedures contained in FAA's Terminal Air Traffic Control Handbook were not clear and resulted in the classification and handling of TWA 514 as a 'nonradar'' arrival. The terms ''radar arrival" and "nonradar. arrival'' were not defined. In view of the available ATC facilities and services and since the flight was receiving radar service in the form of radar monitoring while under the jurisdiction of a radar approach control facility, the procedure should have provided for giving altitude restrictions in an approach clearance for an aircraft operating on an unpublished route prior to its entering a segment of the published approach procedure. The ATC system was deficient in that the procedures were not clear as to the services the controllers were to provide under the circumstances of this flight. The flightcrew believed that the controller would not clear them for an approach until they were clear of all obstructions. The depiction on the profile view of the approach charts neither indicated the position of Round Hill intersection nor did it contain all minimum altitudes associated with the approach procedure. This information was available on the plan view of the approach chart. The captain noticed the minimum altitude associated with the approach segment from Front Royal to Round Hill but he decided that the flight could descend to 1,800 feet without regard for the 3, 400-foot minimum altitude depicted on the chart because he was not on that segment. 20. 21. 22. 23, 24. ~39- The captain of Flight 514 did not question the controller after receiving the approach clearance, regarding the action the flightcrew was expected to take. Another crew that questioned a similar clearance received further instructions and information which resulted in their accepting a radar surveillance approach to Dulles. ) Both military and civil aviation officials for several years had indicated concern regarding a lack of understanding on their part of what the Air Traffic Control procedures and terminology were intended to convey to the pilots.

[ANALYSIS] The pilot kept the controller informed of the aircraft's position and of the pilot's intentions. Typically, during an instrument approach, numerous radio calls were made as the pilot reported his position, altitude, and intentions. With the advent of radar, the controller was able to observe the aircraft in two dimensions -- range and azimuth -- and was able to _ vector flights to arrive over geographical positions. By issuing headings the controller could prevent the tracks of known IFR traffic from converging if the danger of a collision existed. However, it was still necessary for the pilot to advise the controller of the flight's altitude. As experience was gained in the use of radar, a new language was introduced to pilots and controllers and new procedures were instituted to provide for the control of IFR traffic in the terminal area. The controller played a greater role in maneuvering the aircraft by providing headings and altitudes to pilots. As traffic became heavier and aircraft became faster, the controller played a greater role in the movement of the traffic in an effort to provide an uninterrupted flow of traffic to the runway. In an effort to improve his ability to move traffic, he was assigned blocks of airspace and minimum vector altitude information, which was not known to the pilot, to be used in moving traffic off the published approach routes. The advent of the ARTS III radar system and similar systems now provides the controller with information on properly equipped aircraft in three dimensions -- aircraft altitude, range, and azimuth, as well as ground speed. The volume of terminal air traffic has grown to the point that the FAA has frequently found it necessary to divert flights away from published instrument approach routes in order to improve the flow of traffic. In addition, it has become commonplace to clear pilots to descend below the altitudes published on the terminal area charts and instrument approach charts. Pilots in turn have tended to become I more and more dependent on the air traffic controller to control their flight's altitudes, headings, and airspeeds. Concurrent with this _29.- increasing dependency has been (1).a lessened ability to know the type of terrain over which the aircraft is flying, and (2) in some I cases, limited information regarding the position of the aircraft relative to the airport and obstacles on the ground. Controllers are trained in the air traffic control procedures and the terminology associated with IFR navigation. Pilots, on the other hand, are trained in the operation of the aircraft, air traffic control procedures, and terminology essential to safe operation of aircraft in the airspace system. However, as this case demonstrates, imprecise terminology, unresolved differences of opinion, and unnoticed changes in the definitions and procedures can result in an inadequate understanding on the part of one or both of the participants in the air traffic control situation. At the Safety Board's public hearing, FAA witnesses testified that they were not aware that there was any potential misunderstanding on the part of pilots as to the meaning of the term '"'cleared © for the approach, '' in a case where a nonprecision approach is made, . particularly when the clearance was issued a long distance from the airport.

[CONCLUSIONS] BY THE NATIONAL TRANSPORTATION SAFETY BOARD /s/ JOHN H. REED Chairman /s/ LOUIS M. THAYER Member /s/ ISABEL A, BURGESS Member REED, Chairman, THAYER and BURGESS, Members, concurred in the adoption of this report. (BURGESS, Member, concurring statement on page 43.) °* McADAMS and HALEY, Members, dissented. (See page 45 .) /s/ FRANCIS H. McADAM Member /s/ WILLIAM R, HALEY Member November 26, 1975 Member Burgess Concurring: While I fully concur with the majority, I wish to explain more fully my position regarding the primary difference of opinion as expressed dy the dissenting members. In my judgment the reason why TWA flight 514 was nota radar arrival is predicated on the following: , Generally, the ''final approach course"! is a straightline extension of the centerline of the runway. Although it may 'coincide'' with a radial of a VOR located on the runway, a clear distinction must be made between a vector to the final approach course and a vector to such a radial. Although both the center approach controllers are vectoring aircraft to centerline extensions of the runway, they are doing so for different phases of the aircraft's operation, for different purposes, generally at different altitudes. Once an aircraft is vectored to the "final approach course," (the controller must specifically use these words to describe the purpose of the vector) it becomes a radar arrival and remains such as long as it stays on the final approach course and until radar service is terminated. During this time paragraph 1360 of 7110.8C would be applicable. If the aircraft is taken off the final approach course (for such reasons as traffic or a go-around) the aircraft would cease being a radar arrival unless and until given. another vector to the final approach course, Flight 514 was cleared to the 300° radial even though 80 miles out, not the "final approach course." herefore, by definition, Flight 514 was not a radar arrival. The foregoing finding does not absolve the ATC system since by definition Flight 514 was nota radar arrival. During the course of the investigation it became clear that there was an omission in the ATC handbook concerning exactly when radar service is terminated. It was unfortunate that the handbook did not clearly ~ 44 - require the controller to provide altitude restrictions when an aircraft is operating over an unpublished route for which there is no minimum enroute altitude prescribed, while the flight was being handled as a non-radar arrival.

[ANALYSIS] Qualified instrument pilots and air traffic controllers should know and understand beyond equivocation that the coincidence of the inbound course being an extension of the final instrument approach course does not permit ‘descent to altitudes lower than those published for that air space segment unless specifically authorized by ATC. A clear, precise definition of final approach course and final instrument approach course should preclude future misunderstandings. Neither of these terms was defined in the AIM at the time of this accident. However, the AIM glossary did contain a definition of ''Final Approach - IFR" wherein the final instrument approach course is shown to be confined to the final approach segment of the instrument approach procedure and that it begins at the final approach fix. _ The issue of when flights are or are not radar arrivals must also be resolved. It is difficult for a pilot who is operating in a radar environment and communicating with a radar controller to realize that, under some circumstances, his flight is, without formal notification, considered to be a nonradar arrival and subject to a different ATC procedure. Specifically, he may not realize that the responsibility for obstacle clearance shifts from the controller to the pilot under some circumstances without the pilot being specifically informed. While the Safety Board recognizes that the FAA is concerned about radio frequency congestion in busy terminal areas, any control procedure which effects a change in the responsibility for providing terrain clearance must be communicated and clearly understood by both pilots and controllers. If radar service is terminated, the crew should be so informed. Then they will be prepared to resume the responsibility for navigation which was vested in the controller while the flight was classified and handled as a radar arrival. The ARTS III system provides, as previously noted, information capability not formerly available to controllers. The Safety Board has previously recommended that the altitude information capability of this equipment be used as an additional safety factor in the terminal area to help prevent controlled flight into the ground. In the case of Flight 514, the controller testified that he could not clearly see the target associated with the flight until he noted that the altitude was 2,000 feet. Immediately thereafter, he attempted to contact the flight to verify its altitude, but impact had already occurred. The FAA has taken action to install an altitude deviation warning in the ARTS III system which should be beneficial in alerting controllers to altitude deviations in the terminal area. Although the record of this investigation shows that the weather was a factor in the occurrence of the accident, it was not of such nature as to have made the accident inevitable. The icing encountered by the aircraft in the descent was apparently eliminated by the anti-icing systems. The intensity of the turbulence may have been sufficient to make the control of the aircraft somewhat difficult. The excursions of the traces on the flight data recorder are indicative of light to moderate turbulence. The possible effect of the high winds on the indicated altitude has been discussed previously. While the evidence does not indicate whether the crew was aware of the SIGMETS issued for the Washington area, there is no evidence to indicate that knowledge of the SIGMETS would have caused the crew to operate any differently than they did. "= 35 - The CVR indicates that the crew did encounter considerable turbulence during the descent.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] Further, from information available to him on the instrument approach cluit and from his previous experience in the Salt Lake City area, he should have known that 6,000 ft was well below the elevations of surrounding mountains. Therefore, he should have snsisted on absolute certainty about where the flight was to hold. When the approach controller issued the holding instructions, he was not aware that cosounications and, therefore, radar control, later’ would be interrurted. Consequently, the holding instructions were imprecise and containec an ambiguity which the flightcrew failed to detect. 8/7 The GPWS probably functioned becauseI the GPWS ewitch was found in the normal position. Additionally, the rapid increase in the FDR altitude trace and corresponding decfease in the airspeed trace during the final 4 to 5 secs of flight, and the impact attitude of about 15°, indicate that the pilot reacted sharply to:such a stimulus. ~ 30 - The Board has noted this lack of precision in communication in other accidents 9/, and we believe that some of it is attributable to complacency while operating in the radar environment. When under radar control, flightcrew communications and adherence to prescribed procedures may tend toward imprecision because they know that the controller has the means to detect and correct mistakes. On the other hand, the coatroller may be less precise in his communications and adherence to prescribed procedures because he has the seans to correct any mistakes or misunderstandings that might occur. Consequently, after lengthy exposure to the pure radar environment, both flightcrews and air traffic controllers develop habits of imprecision in their communications with each other and in their adherence to prescribed procedures. Further, the exposure can lead to a loss of knowledge of procedures which, generally, were developed for use in the nonradar environment or for use in the event of lost communications and which may be used rarely with precision in the pure radar environment. Flightcrews and controllers alike should consciously strive for precision in their communications with each other and in their adherence to prescribed procedures, not only to avoid events similar to those which led to this accident, but also because the loss of communications between the flightcrew and controller always terminates radar control and prevents both parties from correcting mistakes or clarifying ambiguities. Another problem inherent in situations involving malfunctions of aircraft systems in flight is the division of responsibilities among members of the flightcrew while the malfunction is being resolved. The Safety Board has addressed these responsibilities in a number of accident reports. 10/ In this instance, because of the lack of CVR information, the manner in waich the captain coordinated and managed the activities of the first officer and the second officer is not explicitly known. --- Footnotes: [9/ NTSB-AAR-73-15, North Central Airlines, Inc., and Delta Air Lines, Inc.,]

[ANALYSIS AND CONCLUSIONS > ANALYSIS] He should have furtuer realized that while he was providing radar vectors and radar navigatiunal guidance to an aircraft operating at MYA, he was also required to provide advisories in the event the aircraft deviated from its protected airspace. If the controller was unable to communicate with the flightcrew, he could not provide the deviation advisories tc them. Therefore, in the absence of a request for emergency handling, he should have taken one of the following actions: (1) Directed the flight to a protected area which would not have required the controller's provision of radar navigational guidance, or (2) dented the request to leave the frequency, Notwithstanding the controller's alternatives, he undoubtedly was misled by the captain's suggestion that the flight would only be off frequency “for a little minute." Given the aircraft's position, altitude, and groundspeed at that time (0129:51) and the flight's clearance to "turn right and proceed direct to the Salt Lake VOR...", the controller knew that the flight was safe from obstructing terrain for well over a minute, As the flight progressed, the aircraft passed over the VOR about 0132, or more than 2 min after the captain inplied that the flight would be off the frequency for a short time. d , ; ‘s "ek if eh om li Ng el ee eel lt get eee tes em ek ne ee he ~- 27 - oo ‘ Ir fact, the flight was absent from the controller's frequency for about 7 1/2 min. ‘The ARINC transcripts show that 2 nin 16 sec cf the ? 1/2-min period were consumed in cstabiishing communications with the maintenance controller. Consequently, the Board cannot explain why the captain thought the flight's absence from the frequency would be only “a little minute." However, flightcrew probably was nct concerned with the passage of time Secause they believed themselves in « safe area, and they were intent on solving the ianding gear problem and a difficult electrical system problem. In eny event, the whole pattern of imprecise communications with approach control suggests a somewhat casual and complacent attitude toward management of the flight. * bal I , During the 7 1/2-min period, (about 0136), it became obvious to the controllers that the flight would cross the 331° radia) on a northerly track instead of turning right to intercept the radial and flying inbound on the 331° radial to the VOR. Consequently, the controllers attempted to contact the flight through the Salt Lake City and Ogden VOR's but were not successful because the flight was not monitoring the VOR's for voice transmissions even though both VOR receivers were tuned to the Salt Lake City VOR frequency.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] Soe An, teens Nal, NO my ue ae oy eo. a " 2? a, ‘ : . n nt i my * 7 . . » i : . : - . . i . . . a , . . ’ a ko gee a. I itll le daa 1 . i oo fe ae * , > rs . , : -¥ oe eae ge 5 mr . I * cea IEE NI ene en eS ee Me tren ee me ee nat ena eR NEM ONE Ea ES ee RT Oe EET FPERNR i + ~ 29. part. However, since the MSAW alert was flashing and since the aircraft was headed toward areas where the MVA's were 9,000 ft and higher, the controller should have placed more emphasis on the urgency of the action he told Flight 2860 to take, and he should have given the flight instructions t> immediately turn and. immediately c)imb. The cenditions in the cockpit of Flight 2860 after the flight reported back on approach control frequency are not known because of the lack of CVR information. However, based on weather reports and witness reports, the flight apparently entered instrument flight conditions during the inbound turn, if not before, and the flightcrew was not aware that a dangerous situation was developing. Consequently, the controller's instruccions probably surprised them sufficiently to cause delays in their responses. Additionally, simulation tests indicate that the GPWS would not have provided a warning until 7.7 to 10.2 secs before impact, which because of the rapidly rising terrain was too late. 8 Clearly, 1t was a preventable accident because so rany independent events had to corbine sequentially to produce the accident, and slight alterations in any of these events could have prevented it. However, we conclude that the most critical of the events was the manner in which understanding was reached on the holding clearance, because if the holding clearance had been properly given and properly understood the events that followed either wouid not have affected the safety of the aircreft or would not have occurred. We believe the majcr problem with the holding clearance was the lack of precision in the communications between the varties involved. ; The captain knew that he had only one radio and that he would have to terminate ATC communications, and radar control, in order to ccmmunicate with United's maintenance controller. Further, from information available to him on the instrument approach cluit and from his previous experience in the Salt Lake City area, he should have known that 6,000 ft was well below the elevations of surrounding mountains.

[PROBABLE CAUSE] The aircraft's GPws probably functioned from 7.7 to 10.2 sec before impact but not in time for the flightcrew to prevent the aircraft's collision with terrain which rose at a 32° angle from the horizontal. The accident was not survivable because severe impact forces destroyed the aircraft and subjected the flightcrew Satie tthe tate at a 3.2 Probable Cause The dational Transportation Safety Board determines that the probable cause of this accident was the approach controller's issuance and the flightcrew's acceptance of an incomplete and ambiguous holding clearance in combinacion with the flightcrew's failure to adhere, to prescribed impairment-of-comnunications procedures and prescribe holding procedures, The controller's and flightcrew's actions ate attributed to probable habits of imprecise communication and of inpracise adherence to procedures developed through years of exposure to Operations in a radar environment, Contributing to the accident was the failure of the aircraft's No. 1 electrical system for unknown reasons, 4, RECOMMENDATIONS On April 3, 1978, the Safety Board issued Safety Recommendations A-78-21 and A-78-22 to the Federal Aviation Administration as follows: "Review the adequacy of current cockpit voice recorder preflight testing procedures to assure satisfact ry System operation. (A-78-21) "Review the reliability of cockpit voice recorder units to assure that the mean time between failure is not excessive. {{A~78-22)" BY THE NATIONAL TRANSPORTATION SAFETY 30ARD /s/ JAMES B. KING Chairman /s/ FRANCIS H. McADAMS Member /s/ PHILIP A. HOGUE Menber /s/ ELWOOD T. DRIVER Member July 27, 1978 * Fd t PRRLORRTERE oe NI. SEE SO I Ep re: NR Pr: 0 allie or ates re: 7 * e . ag . e 4 5 bs 7 a Y + =. I I I 5. APPENDIXES APPENDIX A INVESTIGATION AND HEARING l. Investigation The National Transportation Safety Board was notified of the accident about 0220 on December 18, 1977. The Safety Board immediately dispatched an investigative team to the scene. Investigative group were established for operations/witnesses, air traffic control, weather, human factors, Structures, powerplants, systems, flight data recorder, maintenance records, and cockpit voice recorder, Parties to the investigation were: The Federal Aviation Administration, United Airlines, Inc., Air Line Pilots Association, Professional Air Traffic Controllers Organization, Douglas Aircraft Company, International Association of Machinists, Pratt & Whitney Division of United Technologies Corporation, 2. Hearings There was no publis hearing. Depnsitions of material witnesses were taken in Salt Lake City, iltah, February 28, 1978, and San Francisco, California, March 2, 1978.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] Another problem inherent in situations involving malfunctions of aircraft systems in flight is the division of responsibilities among members of the flightcrew while the malfunction is being resolved. The Safety Board has addressed these responsibilities in a number of accident reports. 10/ In this instance, because of the lack of CVR information, the manner in waich the captain coordinated and managed the activities of the first officer and the second officer is not explicitly known. However, it is known from the ATC and ARINC commmications recordings that the captain was actively involved in resolution of the electrical 9/ NTSB-AAR-73-15, North Central Airlines, Inc., and Delta Air Lines, Inc., O'Hare International Airport, Chicago, Illinois, December 20, 1972. NTSB-AAR~75-16, Trans World Airlines, Inc., Berryville, Virginia, December 1, 1974. NTSB-AAR-77-8, Jet Avia, Ltd., Palm Springs, California, January 6, 1977. , ’ 10/ NTSB-AAR-70-14, Scandinavian Airlines System, near Los Angeles, California, January 13, 1969. NTSB~-AAR-73-8, Bolawk \irlines, Inc., Albany, New York, March 3, 1972. ‘NTSB~-AAR-73-14, Eastern Air Lines, inc., Miami, Florida, December 29, 1972. gr - 3] - problem and in obtaining a holding clearance. Consequently, the captain probably was distracted by the electrical problem from supervision of the flying activities, including obtaining the holding clearance and the manner in which the first officer flew the holding pattern, Similarly, it is possible that the first officer was monitoring the resolution of the electrical problem and, therefore, was paying less than full attention 1.0 ATC communications and to flying the aircraft, Since this type of situation is dynamic because the aircraft must be flown while the malfunction is reso.ved, it follows that the captain must manage the flightcrew in a manner which will insure absolute safe operation of the aircraft during the interim. Therefore, although each situation will vary depend‘ng on the type of aircraft involved, the complexity and criticality of the malfunction, the composition of the flightcrew, and many other factors, it remains that the captain's first and foremost responsibility is to insure safe operation of the ai~craft. To achieve this objective, he must relegate other activities accordingly. 3. CONCLUSIONS Findings 1. The flightcres were properly certificated and were qualified for the flight. 2. There was toxicological evidence of alcohol in the second officer's body which according to the weight of medical opinion most likely resulted from his ingestion of alcohol during the 8-hr period preceding the flight; however, since there was no corroborative evidence of alcohol consumption or the effects thereof, the degree of impairment, if any, of the second officer's physiological and mental faculties could not be determined. --- Footnotes: [9/ NTSB-AAR-73-15, North Central Airlines, Inc., and Delta Air Lines, Inc.,]

[FINDINGS] When initially dispatched, the aircraft's No. 1 a.c. electrical generator was inoperative, but repairs were completed and the dispatch release wes revised accordingly before the flight departed Saa Francisco. The aircraft's No. 1 electrical system malfunctioned during the flight's descent for the approach to Salt Lake City airport; the No. 1 electrical bus was inoperative and all of its associated electrical components were inoperative. 7410371 GQ The CRETE ersten ny EERE SA NED AER Ot REAL” BS: SPM ic es acide EEN EE TIN AEE ERC LEI I TS RRO POE REP ne EY Other than components that were powered through the No, 1 electrical bus, there was no evidence of malfunction or failure of the aircr. t's other Systema, including flight insirument and navigational systems, or its structure, powerplants, or fligt<« controls. Contrary to United's DC-8 Flight Handbook, the No. 1 communications radio was powered trough the No. 1 electrical bus; the radio was inoperative after the loss of the No. 1 bus. , The flightcrew was unable to verify lan ing gear extension because the landing gear indicator sysxem was powered through the No. 1 electrical bus. Shortly ufter the flight established cc murications with Salt Lake City approach coatrol, the first officer began flying the aircraft and the captain managed the radio communications. Contrary to regulations, the flightcrew did not infora ATC of the loss of a communications radio, the extent to which the loss impaired the flighc's capability to operate IFR in the ATC system, or the assistance desired from ATC. Because the captain wanted to communicate with United's systen line maintenance control in San Francisco, he requested a holding clearance from the Salt Lake City approach controller. The holding clearance tssued by the approach controller was incomplete and attempts to clarify the clearance resulted in an ambiguity. The approach controller intended that Flight 2869 hold northwest on the 551° radial of the Salt Lake City VOR, but he dtd not specify the radial. The captain apparently intended to hold north of the Salt Lake City VOR but did not request a complete holding clearance, including a holding radial. Because the approach controller did not issue a holdine radia:,and because the captain did not request a holding radial, the first officer assumed the 360° radial to be holding radial. meee. MR - een 19. 20. 21. 22. 23.

[FINDINGS] The Safety Board concludes that the controller did not assign proper priorities to the requirements of his position as defined by paragraph 22 of the Air Traffic Control Handbook. The subject of the telephone call was not urgent and should have been put off until a time when traffic was light or until the controller could have been relieved at the position by other qualified personnel. 3. CONCLUSIONS Findings 1. The crewmembers of each aircraft were properly certificated and qualified for the flight. 2. The aircraft were certificated and maintained in accordance with FAA requirements, There were no atmospheric restrictions which would have prevented the occupants of each aircraft from seeing the other. Each pilot had the responsibility to see and avoid the other. The pilot and passenger of the airplane concentrated their traffie search in an area ahead and to the right of the flight path. The helicopter could have been hidden temporarily from the airplane pilot's view by the rizht windshield post. Each aircraft would have been visible to the crew of the othe* and could have been seen in time to avoid the collision. Bot:. aireraft were operating in the control zone in accordance with established ATC procedures, except the airplane pilot initially did not have the marker beacon receiver on for the ILS approach. The FAA dié not provide training in the use of BRITE radar; therefore, the controller was neither trained nor qualified in its use and he did not use the BRITE radar as an aid. a ATES Ue TENT as Bea a aaa © \ . 7 : ¢ wena SERENE: EA NS mR IPR EL #2 RARE TI NANTES he apr eee: TARA ERNIE RS Pupe FERS i Pee wee . + ‘ ~18- The controller did not issue a safety edvisury to either aircraft with respect to the other. Both aircraft were in radio contact with the local controller. The local controller was distracted from the traffic situation for over 2 minutes by a telephone call concerning administrative matters. The controlter {{nitially disregarded several radio communieations which led to a busy radio frequency and further delayed a position report from the airplane, The absence of a procedure to provide qualified personnel for assistance when supervisory duties interfere with control of traffic allowed the telephone call to be a distraction. 15. The angle of collision was about 45°, with the helicopter in @ 45° right bank and the airplane descending in straight flight. 16. The center of rotation of the helicopter main rotor disk was about 16 feet below the airplane's fuy/..:-e. Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the failure of each flightcrew to see und avoid the other aircraft and the failure of the local controller to perceive the traffic conflict due to the controller's preoccupation with a nonessential administrative telephone call, Contributing to the accident was a delayed pcsition report from the airplane pilot due to his failure to activate his marker beacon receiver and to controller-induced congestien on the radio frequency and an inaccurate position report from the helicopter pilot.

[ANALYSIS] The tape recording and the transcript (see appendix C) clearly show that his attention was diverted from the current traffic Situation. Furing the conversation, both the helicopter and the airplane made their initial calls, neither of which included their position. It is apparent that the controller was distracted, by; the way in which he responded to the helicopter to "stand by." Purther evidence thet he was distracted is the fact that he cleared an aircraft into takeoff position on the runway and then went back to the conversation for another minute. He issued the takeoff clearance for that aircraft only after terminating the telephone conversation, In addition, he did not respond to three other transmissions received while he was on the telephone. 4 The controller acknowledged the airplane's initial call after the telephone call was terminated, about 30 seconds after the transmission. It is probable that, because he would have received a flight date Strip on the airplane, he remembered the initial contact and acknowledged, It also seeins likely that he forgot the helicopter's initial contact because it occurred when his attention was diverted. This is indicated by his failure after the telephone call to acknowledge that initial contact. He did not speak to the helicopter until it again made an initial call nearly 45 seconds after the end of the telephone conversation. Also, during the telephone conversation, he received an initial cal! from Lance N1I919H, including a position, which he failed to acknowledge until it was repeated 25 seconds after the end of the telephone conversation. . When the controller acknowledged the accident airplane's initial contact, he requested a report when the airplane was at the outer marker. The correlation of transcript and radar plot show that at that time the airplane was within one-half mile of the marker beacon and should have been receiving the marker beacon signal. However, based on the passenger's statement, the pilot did not have the marker beacon receiver turned on until the controller made the request, which Is contrary to normal practice. By sn emer ne anna aes SPS ee ONIN Me ined Rpg ie PEE erm T pee LABS fo Meat eg! oghins of purens neve ng fe Ee , ‘ - Me 2 ~ a a . ‘ * ‘ era) ennai oe oan -1§- the time the pilot turned it cn and was receiving the sigr:al, other trensmissions on the radic frequency prevented him from reporting a position until he was clready on the base leg. Some of these transmissions were from the helicopter and Lance N1919H, which the eontroller had ignored earlier during his telephone conversation, Thus, it appears that after terminating his telephone conversation, the controller was catching up with the traffic situation which had gotten ahead of him. If the airplane pilot had turned on the marker beacon receiver when he began the ILS approach, which is normal procedure, he could have reported his position at the marker when the controller made the request.

[PROBABLE CAUSE] The failure of tie »ederal Aviation Administration to train and qualify tower personnel in the use of the BRITE radar display was also a factor. 4, RECOMMENDATIONS As a result of this accident the National Transportation S-fety Board made the following recommendations: ~-to the Federal Aviation Administration: Through pilot training and examination programs, emphasize to pilots the importance of accurate position reporting in communications with air traffic control factlities. (Class Il, Priority Action) (A-82-58) Revise the helicopter routes contained in the Teterboro Letter to Airmen 81-2 to provide improved separation and thereby minimize the potential for conflicts between helicopters and fixed-wing aircraft traffic. (Class 1, Priority Action) (A--82-59) Provide all pertinent personnel working traffic at BRITE-equipped, nonradar control towers with the proper training and certification ear the use of that equipment. (Class ll, Priority Action) A-~82-80 -~-to the Aircraft Owners and Pilots Association, the National Association of Flight Instructors, the Commuter Airline Association of America, the Helicopter Associaticn International, and the National Business Aircraft Association, Ine.: Through appropriate educational programs and communicaticns, em, hasize to pilots the importance of accurate position reporting in communications with air traffic control facilities. (Class NI, Priority Action) (A-82-61) BY THE NATIONAL TRANSPORTATION SAFETY BOARD /s/ JIM BURNETT Chairman /s/ PATRICIA A, GOLDMAN Viee Chairrn, /s/ FRANCIS H. McADAMS Members G. H. PATRICK BURSLEY ember May 18, 1982 5. APPENDIXFS APPENDIX A INVESTIGATION 1. Investigation The Safety Board was notified of the accident avout 9930 on September 23, 1981, by the FAA's Washington Command Center. An investigator was dispatched immediately to the accident site from the Roard's New York Field Offiee, An investigation team was dispatched from the Board's Washington headquarters with operations, air traffic control, and structures groups. Parties to the investigation included the Federal Aviation Administi ation, Seminole Air Charter, and Ronson Aviation. 2. Public hearing There was no public hearing. Preceding page blank 1 AA MR A, lg MATE DN IE he EO NS aa EN A le in a APPENDIX B PERSONNEL INFORMATION N27470 Captain John D.

[ANALYSIS] The Safety Board also believes that the FAA should emphasize to all pilots the importance and necessity of accurate position reporting. Because of the controller's interpretation of "coming up on,” it is likely that he perceived the helicopter tc be nearly a mile nearer the airport than was actually the case. Although the controller never observed the helicopter, and so advised its crew, he believed the airplane was still outside the outer marker, and he interpreted the helicopter's reported position to be nearly over the stadium. He then concluded that there was no conflict and no need to issue a traffic advisory to either aircraft, Therefore, because of a relatively vague position report from the helicopter, the absence of a timely position report from the airplane, and the earlier distraction of the controller bv an administrative chore, the Safety Board concludes that the controller had an erroneous perception of the relative positions of the two aircraft and therefore did not consider them to be in cotential conflict. Although the helicopter was not precisely following the inbound route "Whiskey" from the southwest as defined by Letter to Airmen 81-2, it was less than 1/2 mile east of the track. The Safety Board does not consider this to be a facte= in the accident since even if the helicopter had been on the track, there would have existed a potential conflict between the two aircraft. The route for helicopters inbound from the south and southeast, designated "Sierra," completely avoids the traffic flow for runway 1. : ve ES : e . PY - 4 . =O TAT vot TPR OT The Safety Board believes that if this route had been emphasized in the Letter to Alrmen as the route for all helicopters inbound from the southwest through southeast, the potential for conflicts such as occurred in this accident would have been minimized, The investigation revealed that although the BRITE radar display had been in the Teterboro tower over 1 1/2 years, no personnel were certified in its use, but it was referred to occasionally. Although it was available and turned on at the time of the accident, it was not being used by the controller. The Safety Board coneludes that if the controller had been certified to use the BRITE display and had used it to rapidly update himself on the traffic situation following the distraction of the telephone eall, he might have perceived the developing conflict and issued an appropriate advisory. Because the Teterboro tower had been understaffed even before the controller's strike, the tower chief had regularly worked in the different control positions on a regular basis and was therefore qualified and current as a controller, However, the need to perform administrative functions in his role as facility chief without backup supervisory capability in this instance led to his distraction from the control of traffic. The Safety Board concludes that the controller did not assign proper priorities to the requirements of his position as defined by paragraph 22 of the Air Traffic Control Handbook. The subject of the telephone call was not urgent and should have been put off until a time when traffic was light or until the controller could have been relieved at the position by other qualified personnel. 3. CONCLUSIONS

[(a) FINDINGS] Their failure to follow these procedures led directly to their failure to properly report their position to ATC when they were over the FAF, énd led to ** ‘r need to execute a missed approach. The Airwest 5 flightcrew's failure to report their position ovec the FAF compromised separation between their aircraft and Northwest 603. The procedures adopted by Fairchild RAPCON and Spokane Tower after the incident will, if adhered to, provide positive separation between aircraft arriving and departing Spokane Airport. (b) Probable Cause The National Transportation Safety Board determines that the probable cause of this incident was the inadequacy of the local air traffic control prccedures to insure positive and adequate separation between I arriving and departing aircraft. Contributing to the incident was the failure of the local controller to recognize and resoive the impending conflict in accordance with the basic mandate to insure positive separation between aircraft. Also contributing to the incident was the fail. ce of the crew of Airwest 5 to . llow company ILS approach procedures and the recommended FAA position reporting procedures. 3. RECOMMENDATIONS As a result of this incident, the National Transportation Safety Board made the following recommendations to the Federal Aviation Adninistratior: Revise the Airman's Information Manual so that the aviation community w‘ll not be misled regarding radar approach control services at locations where the tower cab is not radar equipped and the approach centrol facility has limited, low-altitude radar coverage capability, (Class Il--Priority followup.) (A-76-91.) Review all local departure and arrival procedures and assure that they provide positive separation between aircraft whenever radar and nonradar operations interface. (Class II--Priority followup.) (A-76-92.) La, TRS air! acta sang sie: BY THE NATIONAL TRANSPORTATION SAFETY BOARD /a/ WEBSTER B. TODD, JR. _ Chairman /s/ KAY BAILEY Vice Chairman /s/ FRANCIS H. McADAMS Merber /s/ PHILIP A. HOGUE Member /s/ WILLIAM k. HALEY Menvber August 18, 1976 SF IIE Mee DP Ce ee a ian te oe te ee eee FEES BE PE PS eR ES ET DLT ee APPENDIX A Investigation and Hearing 1. Investigation The National Transportation Safety Board was notified of the incident at 1000 P.s.t., on April 1, 1976. Safety Soard investigators proceeded immediately to Spokane, Washington.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] However, according to the rederal Aviation Regulations, the reyort is not mandatory unless specified by ATC, and, in this instance, ATC (Fairchild RAPCON arrival controller) did not tell Airwest 5 to report when over the FAP. Moreover, the separation procedures did not provide for the possibility of a communications fatlure. The separation procedures also specified the use of a separation fix that was less than 4 miles from the airport, which did not meet the requirements of paragraph 940, FAA Terminal Air Traffic Control Manual, 7110.8D. Also, the procedures provided that "Spokane Tower may release an IFR departure until an arrival has or is reported over the final approach fix....'' The tower controllers apparently interpreted "release" to mean "clearing an aircraft for takeoff,'' whereas paragraph 940 required that the departing aircraft: takeofi before the arriving aircraft leaves the fix. These two factors clearly reduced the amount of separation between aircraft arriving and departing runway 21 at Spokane Airport. The ATC procedures used for separation of aircraft in this instance might have been adequate had Spokane Tower been equipped with radar. However, without radar, the local controller's only means of providing separation were visual observation of the two aircraft and distance based on the arriving pilot's report. over the FAF. The local controller apparently assumed that Airwest 5 would be in a position to land because he told the flight to continue after it had reported “inside the outer marker." A controller's primary responsibility is to separate traffic regardiess of any deficiencies in local ATC procedures. When Airwest 5 reported faside the outer marker, the local co: ‘roller should not have assumed that almost 3.9 miles spacing existed between the two airplanes. Instead, he should have realized that the minimum spacing requirements had already been compromised. This realization should have prompted him to direct Airwesi 5 to discontinue the approach and to stop Northwest 603 on the ruaway, since the DC-10 had not yet begun its takeoff roll. Recause he cculd not see either aircraft, he had to rely on Fairchild RAPCON to provide separation directions. However, because of the radar coverage limitations, the RAPCON controller could not identify the two aircraft in time to prevent the conflict. Although the captain of Airwest 5 was apparently justified in assuming that his aircraft was under radar control throughout the approach and missed approach because no information to the contrary was provided by ATC, his failure to inform Spokane Tower of his position when his aircr:€f. was over the OM compromised separation between his aircraft and Northwest 603.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] Although the captain of Airwest 5 was apparently justified in assuming that his aircraft was under radar control throughout the approach and missed approach because no information to the contrary was provided by ATC, his failure to inform Spokane Tower of his position when his aircr:€f. was over the OM compromised separation between his aircraft and Northwest 603. In conformity with good operat:ing practices, Airwest's Ce ” EMORY PURI FO ore te PERI og ete te eee ngs eet a nga tap tube Weiphen “ce egw dee bape RA ONS EES ROTO BREMEN BE NTR ES TIED procedures suggested that he make the report even though Federal Aviation Regulations did not require that he do so. Moreover, its apparent that the manner in which he flew the approach did not corform to Airwest's requirements and led directly to his failure to pruperly report his position and to his need to fly a missed approach rather than land. ; i }} i : The captain of Airwest 5 was at a disadvantage throughout the descent into Spokane because he had to maintain comparatively high thrust levels to operate anti~icing equipment. However, he could have selected several different courses of action to insure that he complied with company procedures. For instance, he could have extended the speed brakes sooner, or he could have requested either a 360-degree turn or delaying vectors to provide the addivional time needed to properly descend to 4,000 feet, to properly configure the aircraft, and to complete the required checks before he intercepted the ILS glideslope. Had he done so, he would have had no difficulty in properly reporting his position or in landing from the approach -- either of which would have enabled ATC to provide adequate separation between his aircraft and Northwest 603. The above-mentioned deficiencies notwithstanding, the Safety Board believes that separation betw-en arriving and departing aircraft cannot be based solely on a nonma’ cory repert from the arriving pilot that his aircraft 1s over the FAF and inbound to the airport, or on essumptions that a pilot flying an instrument approach in instrument meteorological conditions will succeed in landing from the approach. More positive measures must be used because of the possibility of communications failvres or congestion and because of the many factors that can cause the pilot of an arriving aircraft to fly a missed approach. The Safety Board believes that the requirements of paragraph 3b of the Fairchild RAPCON/‘pokane Tower order of April 8, 1976, will preclude the repetition of an incident such as this because a departing aircraft cannot be releasec for takeoff until the position of the arriving aircraft is known. Moreover, the departing aircraft must have departed before the arriving afrcrafe reaches the FAF which should provide the required 4-mile separation. However, the Safety Board believes that arriving flights also should be informed that radar service is terminated when Fairchild RAPCON transfers control of the flight to Spokane Tower.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] Moreover, the departing aircraft must have departed before the arriving afrcrafe reaches the FAF which should provide the required 4-mile separation. However, the Safety Board believes that arriving flights also should be informed that radar service is terminated when Fairchild RAPCON transfers control of the flight to Spokane Tower. Adiitionally, the Safety Board is concerned that similar facilities might exist within the ATC system; that is, a radar approach control which provides service to control towers that are not equipped with radar. Although the GENOT issued by the FAA on April 7, 1976, recognizes this problem, we believe that action should be taken to insure that the proper procedures are employed at all of these facilities. Conclusions (a) Findings 1. There was no malfunction or failure of anv communication or navigation equipment. 2. The flightcrews and air traffic controllers were properly certificated and qualified for the duties they were performing. The Federal Aviation Regulations do not require that the pilot of an arriving aircraft report’ his position when over the FAF to that airport unless specified to do so by ATC. The separation procedures used by Fairchild RAPCON and Spokane Tower did not require the pilot of an arriving aircraft to report his position when over the FAF inbound to the Spokane Airport while on an instrument approach. The separation procedures used by Fairchild RAPCON and Spckane Tower did not provide positive separation betwee: arriving and departing aircraft because too much reliance was placed on a nonmandatory report from the arriving pilot that his aircraft was over the FAF inbound to the airport. Also, the procedures did not provide for at least a 4-mile separation between the arriving aircraft and the departing aircraft. The Spokane Tower local controller did not know positively the pasition of Airwest 5 when he cleared Northwest 603 for takeoff. Following Airw 3st 5's report "inside the outer narker,"” the local controller attempted to apply IFR separatio. criteria on the assumption that Airwest 5 would be able to .and from the approach. The local contruller did not take positive steps at the first indication that the separacion between the two airplanes was guestionable and while he still had the opportunity to delay the takeoff of Northwest 603. Airwest procedures recommended that Airwest flightcrews report their aircraft's position to ATC when it was over the FAF and inbound to the airport on an instrument approach; the flightcrew of Airwest 5 did not follow this recommendacion, The flightcrew of Airwest 5 did not comply with company procedures for flying a normal ILS approach or for oxecuting a missed approach. Their failure to follow these procedures led directly to their failure to properly report their position to ATC when they were over the FAF, énd led to ** ‘r need to execute a missed approach. The Airwest 5 flightcrew's failure to report their position ovec the FAF compromised separation between their aircraft and Northwest 603.

[ANALYSIS] However, the sighting alerted him so that, when the controller issued the clearance, he was realy to execute the evasive maneuver with the neceasary urgency. The circumstances of this accident indicate that automation technology can lead to complacency when it takes the controller "out of the loop" by reducing the need for his interaction with a flightcrew and deemphasizing the cooperative aspects of the air traffic system. Had the radar controller been working with the broad-band radar, he would have ‘been forced to take positive steps to insure separation as soon as American 182 was handed off to him, Of the several steps he could have taken, we mention only two: (1) He could have stopped American 182's climb at FL 331, or (2) he could have asked the flight to report at FL 310 or 330. However, the automatic altitude readouts on the flight's alpha-numeric block induced him to rely solely on his own observation of the PVD data. He did not consider the possibility that he might become distracted or that the computer might fail, and thereby deprive him of his direct readout capability. The Safety Board is concerned that despite the advantages of I narrow-band radar, the ATC system failed to provide the intended safeguards and endangered the livea of 319 persons’. Advances in technology do not necessarily insure greater reliability and safety. The new conflictalert system can serve its intended purpose only when it is not treated as a substitute for timely, positive separation measures which continue to protect air traffic even when the computer fails. Based on the high percentage of human failures in the ATC system, the Safety Board believes that, as long as the human element is part of the total system, an individual's level of competence, the quality of his performance, and his understanding of his primary responstbilities must be given as much managerial sctention ae the equipment he operates. The serious injuries sustained by the passengers were the result of their not having their seatbelts fastered, or properly fastened, although the seatbelt siga was on. Therefore, this accident is another reminder to encourage passengers to keep their seatbelts fastened, not only when the seatbelt sign is on but also when it is off and flight conditions are smooth. Conclusions (a) Findings 1. American 182 and TWA 37 were operating under control of the Wayne sector of the Cleveland Center. . _- . - & ame ~ . wk nny ee ME = — weg -~ 14 - Both flights were on the same jet route and approaching each other head-on; TWA 37 was maintaining FL 350, # erican 182 was cleared to climb through FL 350 to FL 370. The radar controller was aware that a potential traffic conflict axisted between the two flights but assumed that the required separation would exist when the two aircraft passed zach other. The radar controller intended to provide separation if the anticipated separation between the two flights did not materialize. The radar controller became preoccupied with secondary duties and failed to see the impending traffic conflict displayed on his radurscope.

[(a) FINDINGS] The radar controller was aware that a potential traffic conflict axisted between the two flights but assumed that the required separation would exist when the two aircraft passed zach other. The radar controller intended to provide separation if the anticipated separation between the two flights did not materialize. The radar controller became preoccupied with secondary duties and failed to see the impending traffic conflict displayed on his radurscope. About 1 minute before the near collision, the radar controller wasrelieved and he failed to brief the relieving controller adequately. Both controllers failed to notice the unresclived conflict during the transfer of duties. About 50 seconds after taking over the position, the second cortroller detected the conflict and cieared American 182 to escend immediately to FL 330, The two aircraft came within 100 feet of each other. As a result of the abrupt evasive saneuver, 24 occupants of the aircraft were injured, 3 of them seriously; the latter injuries were associated with failure to make proper use of the seatbelt. (b) Probable Cause The National Transportation Safety Board determines that the probable cause of this near-collision was the failure of the radar controller to apply prescribed separation criteria when he first became aware of a potential traffic conflict,which necessitated an abrupt collision avoidance maneuver. He also allowed secondary duties to interfere with the timely detection of the impending traffic conflict when it was displayed clearly on his radarscope. Contributing to the accident was an incomplete sector briefing during the change of controller personnel--about 1 minute before the accident. ~ 15 - po 5 BY THE NATIONAL TRANSPORTATION SAFETY BOARD: Acting Chairman /s/ FRANCIS H. McADAMS Member /s/ LOUIS M. THAYER Member /s/ ISABEL A. BURGESS Member /s/ WILLIAM R. HALEY Member January 28, 1975 -~l17- APPENDIX A INVESTIGATION AND HEARING 1. investigation The Safety Board was notified of the accident at 2000 e.s.t. on November 26; 1975. Investigators proceeded {{mmediately to Cleveland, Ohio, Detroit, Michigan, New York, New Yorx, and Los Angeles, California. Parties to the investigation included the Federal Aviation Administration, American Airlines, Inc., Allied Pilots Assoclation, Air Line Pilots Association, Trans World Airlines, Inc., and the Professional Air Traffic Controllers Organization. 2. Hearing There was no public hearing. Depositiona were taken on December 12, 1975. APPENDIX B CREW AND CONTROLLER iNFORMATION Captain Guy Eby (American Airlines) Captain Eby, 47, holds Airline Transport Pilot Certificate No. 261304 with type ratings in DC-3,6,7,10, L-188, CV~-240, 340, 440,880,990 and B-707,720. At the time of the accident he had accumulated about 21,600 flight-hours, 670 of which had been in the DC-10.

[ANALYSIS] Although the radar controller was aware of a potential conflict, he assumed that American 182 would have climbed to FL 370 before passing TWA 37, which was cruising at FL 350. Ih. addition, he assumed that, by keeping an eye on the situation,he would be able to take timely steps if the anticipated separation did not materialize. Both of these assumptions were not compatible with safe and positive traffic control practices and procedures. By the time the radar controller's first assumption was invalidated, his second assumption, intended as a safeguard, did not work as planned because other activities distracted him. The fact that he consented to be relieved from his position about 2 minutes before the near-collision proves that he had becotie preoccupied with secondary duties to the extent that lie had failed to see the impending conflict that was clearly displayed on his radarscope by that time. The Safety Board belteves that the principle lesson in this near disaster is that intent to separate traffic can never be a substitute for positive action at the first opportunity to insure separation. During the briefing associated with the transfer of duties to radar controlier No. 2, the first controller did not mention American 182, undoubtedly because he was no longer thinking about the flight as an unresolved problem. Since radar controller No. 2 had no reagon to expect. that the responsibility he accepted included an acute problem, it is fortunate that he noticed the problem within 50 seconds after taking over the position. However, this timely discovery does not exonerate both controllers from their failure to notice the conflict during the transfer of duties. The Safety Board concludes that the briefing was incomplete because neither controller reviewed the actual situation as depicted on the PVD, The general discussion about the cloud tops and other traffic that took place on the Wayne sector frequency probably prompted the captain of American 182 to look outside and observe the weather. His remark, "There he is", 1 second before the controller issued the descent clearance, was undoubtedly prompted by aircraft lights he saw. Although the captain's recollection is vague, his remark probably referred to the presumed sighting of the aircraft that, according to a prior statement by the controller, was flying at FL 370. Considering the darkness, the climbing attitude of his aircraft, the restricted visibility conditions, the high altitude, and the closing speed, it would have been difficult for the captain to determine if a traffic conflict existed and, if so, what corrective action to take when he first sighted the lights. However, the sighting alerted him so that, when the controller issued the clearance, he was realy to execute the evasive maneuver with the neceasary urgency. The circumstances of this accident indicate that automation technology can lead to complacency when it takes the controller "out of the loop" by reducing the need for his interaction with a flightcrew and deemphasizing the cooperative aspects of the air traffic system.

[PROBABLE CAUSE] Also causal was the diminished ability of the accident pilots to see and avoid the other aircraft due to flight path geometry, out-the-window view obscuration by aircraft structures, attention demands associated with the air show performance, and the inherent limitations of human performance that can make it difficult to see another aircraft. Contributing to the accident were the lack of Federal Aviation Administration (FAA) guidance for air bosses and air show event organizers on developing plans and performing risk assessments that ensure the separation of aircraft that are not part of an approved maneuvers package and the lack of FAA requirements and guidance for recurrent evaluations of air bosses and direct surveillance of their performance. Aviation Investigation Report AIR-24-07 77 4. Recommendations 4.1 New Recommendations As a result of this investigation, the National Transportation Safety Board makes the following new safety recommendations.

[ANALYSIS] Further, the de Havilland DHC-3 pilot’s ability to see and avoid the de Havilland DHC-2 was limited, in part, due to the little apparent motion of the de Havilland DHC-2 as the two airplanes converged at a relatively constant angle for about 3 minutes before the collision. The report noted that the lack of apparent motion reduced the likelihood that the de Havilland DHC-3 pilot would see the other airplane in his periphery, citing a study stating that human ambient vision (a component of the human visual system that processes information in a peripheral field of view) is more sensitive to motion than to fine detail (NTSB 2021b, 33). Aviation Investigation Report AIR-24-07 60 operating an aircraft so as to see and avoid other aircraft….Pilot[s]…may not pass over, under, or ahead of [another aircraft] unless well clear.” Likewise, the performers trust that the air boss will ensure necessary coordination between pilots, communicate clearly and concisely, maintain situation awareness, provide positive control, and apply good judgment.54 This relationship is similar to that of an air traffic controller and a pilot conducting flight under visual flight rules. In the air traffic control environment, although it is incumbent upon pilots to remain vigilant about their surroundings as the rules of operational right-of-way dictate, the controller shares responsibility to ensure separation from other aircraft when issuing instructions to that pilot.55 While under the direction of an air boss, a performer reasonably expects that the air boss would not issue directions that would put aircraft in conflict, or that the air boss would issue corrective instructions if a performer unintentionally maneuvers an aircraft into a position where a conflict might occur. According to the Wings Over Dallas air boss, his intent for the next pass was for the fighter formation (which was composed of faster airplanes than the bomber group) to pass the bomber group off the bombers’ left side, such that he expected the fighters to be past all of the bombers before transitioning to their assigned 500-ft show line and for all performers to maintain visual separation. However, the position data for the airplanes showed that the Bell P-63F was not yet past the lead bomber when the Bell P-63F began to move toward the assigned 500-ft show line, on a converging path with the Boeing B-17G.

[ANALYSIS] 2. Analysis 2.1 Introduction The accident occurred when two CAF-operated warbird airplanes, a Boeing B-17G bomber and a Bell P-63F fighter, collided in flight during an air boss– directed warbird performance that included multiple, dissimilar aircraft at the CAF’s Wings Over Dallas air show. The analysis discusses the accident sequence and evaluates the following safety issues: • The factors that limited the ability of the Boeing B-17G pilot and the Bell P-63F pilot to see and avoid each other’s aircraft, and the inherent limitations of the see-and-avoid concept for collision avoidance (section 2.2.1). • The air boss’s ineffective aircraft deconfliction strategy for the accident performance (section 2.2.2). • The lack of adequate air show safety oversight, including the requirements for the contents of the air show Participants Safety Briefing (section 2.3.1), the need for administrative controls and documented safety risk assessments (section 2.3.2), and the need for air boss oversight, including recurrent evaluations, standardized communications, and FAA surveillance (section 2.3.3). • Air show safety culture issues, including evidence that CAF pilots did not report observed safety concerns, and the limited ability of the ICAS to influence culture in an industry composed of various operators, individual performers, and individual air bosses (section 2.4). Having completed a comprehensive review of the circumstances that led to the accident, the investigation determined that none of the following were factors: • Flight crew qualifications. The Boeing B-17G pilot, copilot, and flight engineer and the Bell P-63F pilot were certificated in accordance with federal regulations and were current and qualified in accordance with CAF requirements. • Airplane mechanical condition. Maintenance records for both airplanes indicated that each was maintained and inspected in accordance with its respective applicable maintenance requirements. Photographic and video evidence that captured imagery of both airplanes during the performance showed that both were intact and maneuvering in a manner consistent with controlled flight before the collision. Examination of the wreckage of both Aviation Investigation Report AIR-24-07 54 airplanes identified no evidence of any precollision anomaly or failure that would have precluded their normal operation. • Flight crew medical fitness. The Boeing B-17G pilot, copilot, and flight engineer and the Bell P-63F pilot possessed valid and current FAA medical certificates appropriate for the flight operations. The NTSB reviewed their FAA medical certification information and the autopsy and toxicology reports for all personnel on board both airplanes. Considering the autopsies’ limitations (due to the severity of some injuries) and the operational circumstances of the accident, our evaluation determined the following: o The Boeing B-17G pilot and the Bell P-63F pilot each had coronary artery disease.

[ANALYSIS] The organization has no real means to proactively identify or assess risk in the air show environment, and its lack of authority over air shows and air bosses hampers its ability to determine or affect air show safety culture. Thus, the NTSB concludes that, although ICAS was responsible for administering the FAA-accepted program that specified the requirements for a Aviation Investigation Report AIR-24-07 73 person to obtain an air boss LOA, its role did not involve actively monitoring air shows or the performance of air bosses, underscoring the need for the FAA to address the regulatory void related to recurrent evaluations of air bosses and direct surveillance of their performance at air shows. Safety Recommendations A-24-31 through -34 recommend that the FAA address these issues. Aviation Investigation Report AIR-24-07 74 3. Conclusions 3.1 Findings 1. No pilot qualification deficiency or airplane malfunction or failure was identified, and there is no evidence that any flight crewmember’s medical condition or use of medications contributed to the accident. 2. Although the fighter lead pilot was confused by the air boss’s unclear directives, the fighter lead pilot and the position 2 fighter pilot visually ensured separation from the bomber group airplanes when passing and crossing in front of the Boeing B-17G before mistakenly aligning with the incorrect show line. 3. The ability of the Bell P-63F and the Boeing B-17G pilots to see and avoid each other was limited due to flight path geometry, out-the-window view obscuration by aircraft structures, the limitations of human performance that can make it difficult to see another aircraft, and the attention demands associated with maintaining flight along the assigned show lines and, for the Bell P-63F pilot, as a trailing aircraft in a formation. 4. The circumstances of this accident underscore the inherent limitations of the see-and-avoid concept for collision avoidance. 5. The air boss’s deconfliction strategy for the accident performance, which relied on his real-time, predictive assessment of airplane locations and the ability of the pilots to see and avoid other airplanes, was ineffective because the flight paths of the Boeing B-17G and the Bell P-63F converged as each pilot maneuvered toward their respective assigned show lines. 6. Federal Aviation Administration and International Council of Air Shows Inc. guidance did not adequately address the need to better mitigate the collision risks associated with air boss–directed performances involving multiple, dissimilar aircraft. 7. Compared to an approved maneuvers package, an air boss–directed performance involving multiple, dissimilar aircraft represents an increased workload for both the air boss and the pilots, as the air boss bears the cognitive load of having to create the performance sequence in real time and then issue directives, and the pilots must anticipate, understand, and comply with the directives while maintaining visual separation from other aircraft.

[ANALYSIS] Both airplanes were operating as air tours in a geographic area known for its high concentration of air tour traffic, and our investigation cited the inherent limitations of the see-and-avoid concept as directly causal to the accident.53 Thus, the NTSB concludes that the circumstances of this accident underscore the inherent limitations of the see-and-avoid concept for collision avoidance. 2.2.2 Ineffective Aircraft Deconfliction Strategy By definition, an air boss has primary responsibility for the control of air show operations and is expected to maintain situation awareness, ensure necessary coordination between pilots, communicate clearly and concisely, and provide positive control of the air show operations. According to the Wings Over Dallas air boss, he typically accomplished aircraft separation through visual, lateral, timing, and altitude deconflictions. He described these processes as situation-dependent with many variables that required flexibility. For example, he could direct one performer to follow another performer’s aircraft (visual deconfliction), assign performers different show lines (lateral deconfliction), have a performer delay entering the flying display area until after another airplane lands (timing deconfliction), or assign aircraft to fly at different altitudes. Within the air show environment, there is mutual trust between the air boss and the performers. The air boss trusts that the performers will follow directives, maintain situation awareness, and ensure adequate separation between their aircraft and others, as required by 14 CFR 91.111(a) and 91.113(b), which specify, respectively, that “no person may operate an aircraft so close to another aircraft as to create a collision hazard” and that “vigilance shall be maintained by each person 53 In the Ketchikan accident report, the NTSB determined that the de Havilland DHC-2 pilot had no opportunity to see and avoid the other airplane because it was obscured by aircraft structure. Further, the de Havilland DHC-3 pilot’s ability to see and avoid the de Havilland DHC-2 was limited, in part, due to the little apparent motion of the de Havilland DHC-2 as the two airplanes converged at a relatively constant angle for about 3 minutes before the collision.

[ANALYSIS] However, the position data for the airplanes showed that the Bell P-63F was not yet past the lead bomber when the Bell P-63F began to move toward the assigned 500-ft show line, on a converging path with the Boeing B-17G. Thus, the NTSB concludes that the air boss’s deconfliction strategy for the accident performance, which relied on his real-time, predictive assessment of airplane locations and the ability of the pilots to see and avoid other airplanes, was ineffective because the flight paths of the Boeing B-17G and the Bell P-63F converged as each pilot maneuvered toward their respective assigned show lines. 2.3 Air Show Safety Oversight CAF pilots participated in various air shows each year, some of which were hosted by other event organizers, held at locations other than an aircraft’s base, or included performances that combined CAF-operated aircraft with warbirds operated 54 The ICAS ABRP evaluation form listed each of these as air boss responsibilities. 55 FAA guidance related to pilot and controller roles and responsibilities stated that “to maintain a safe and efficient air traffic system, it is necessary that each party fulfill their responsibilities to the fullest. The responsibilities of the pilot and the controller intentionally overlap in many areas providing a degree of redundancy” (FAA 2024a, 5.5.1). Aviation Investigation Report AIR-24-07 61 by other museums or private owners. According to experienced CAF air show pilots, warbird performances commonly involved multiple solo airplanes, including a mix of bombers and fighters, and such performances were not particularly complicated to fly. However, one pilot said that, typically, the fighters would remain at higher altitudes than the bombers, and the altitudes would be discussed ahead of time with the air boss, during the participants safety briefing. 2.3.1 Participants Safety Briefing Requirements The air boss briefed general aspects of the accident performance with the participants during the required Air Show Participants Safety briefing. According to the FAA IIC for Wings Over Dallas, the air boss’s briefing addressed all the required items listed in the FAA Order 8900.1 guidance for FAA inspector surveillance of air shows. Further, none of the briefing attendees interviewed postaccident, which included the FAA IIC, the FAA inspector trainee, the observer air boss, CAF management personnel, and CAF performers, voiced any concern about the air boss’s briefed plan for the accident performance. Further, the FAA certificate of waiver for the air show, FAA Order 8900.1 guidance, and the ICAS ABRP manual did not specify any required procedures for maintaining separation between multiple, dissimilar aircraft operating within the same flying display area and not part of an approved maneuvers package, as was the case with the accident performance. Also, none of these materials specified a requirement for air show event organizers or air bosses to perform a documented risk assessment for each air show performance.

[CONCLUSIONS > FINDINGS] Aviation Investigation Report AIR-24-07 75 8. A lack of administrative controls, such as a prebriefed aircraft separation plan with defined lateral, altitude, and timing deconflictions, directly contributed to the in-flight collision. 9. A documented safety risk assessment for each performance would allow air show event organizers and air bosses to identify hazards and determine and apply effective mitigations, such as establishing administrative controls to ensure aircraft deconfliction for performances involving multiple, dissimilar aircraft. 10. Recurrent evaluations of air bosses as part of the letter of authorization renewal process would provide objective and ongoing assessments of their performance to help ensure the safety of air show operations. 11. Standardized phraseology for references to show lines and other safety considerations that avoids the use of ambiguous terms could help prevent confusion and ensure the clarity and brevity of air boss–provided directives while minimizing radio congestion. 12. The lack of guidance and required surveillance tasks for Federal Aviation Administration inspectors assigned to air shows related to the direct observation of an air boss’s performance represents a missed opportunity for inspectors to detect performance-related safety issues and provide debriefing feedback to address these issues, such as unclear directives, inadequate aircraft deconfliction strategies, or deviations from a previously briefed plan. 13. The Commemorative Air Force’s lack of a strong, clearly defined safety risk assessment plan resulted in air show production decisions that were not systematically developed to determine acceptable levels of risk and were susceptible to influences unrelated to safety, including pressures to deviate from the intended limitations on when revenue passenger rides could be conducted during air show operations. 14. Although the International Council of Air Shows Inc. was responsible for administering the Federal Aviation Administration (FAA)-accepted program that specified the requirements for a person to obtain an air boss letter of authorization, its role did not involve actively monitoring air shows or the performance of air bosses, underscoring the need for the FAA to address the regulatory void related to recurrent evaluations of air bosses and direct surveillance of their performance at air shows. Aviation Investigation Report AIR-24-07 76 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of the accident was the air boss’s and air show event organizer’s lack of an adequate, prebriefed aircraft separation plan for the air show performance, relying instead on the air boss’s real-time deconfliction directives and the see-and-avoid strategy for collision avoidance, which allowed for the loss of separation between the Boeing B-17G and the Bell P-63F airplanes.

[ANALYSIS] However, at no time before or during the accident performance did he set designated altitudes or airspeeds for the pilots to fly, and the directive he issued for the fighter formation to cross laterally in front of the bomber group relied entirely on his real-time, predictive assessment that the fighter formation would arrive at the show line first and that the pilots would see and avoid each other. As discussed in the previous section, this strategy presents a higher workload in terms of communication and information processing and an increased demand for situation awareness for both the air boss and each pilot. Thus, the NTSB concludes that a lack of administrative controls, such as a prebriefed aircraft separation plan with defined lateral, altitude, and timing deconflictions directly contributed to the in-flight collision. The NTSB notes that after this accident, on July 15, 2024, the FAA issued SAFO 24005, “Mass Aircraft Demonstrations at Aviation Events,” to recommend proactive risk mitigation strategies for the pilots, air bosses, and event organizers of civilian air shows that include mass aircraft demonstrations, particularly those involving multiple, dissimilar aircraft that are not part of an approved maneuvers package (FAA 2024c). The SAFO recommended distributing to the pilots a detailed written plan for the performance; conducting mandatory preflight briefings and postflight debriefings that review all aspects of normal and emergency procedures; using aircraft deconfliction strategies characterized by complete geographical, lateral, and time separation; and flying simple racetrack patterns to avoid complex maneuvering and loss of visual separation. Aviation Investigation Report AIR-24-07 64 The SAFO provided a link to an ICAS best practices document that contained guidance for drafting and implementing a performance plan, recommended preflight briefings and postflight debriefings, and included examples of a separation strategy that has proven effective for the event organizer of a recent large air show that featured multiple warbirds. However, it did not include guidance for developing effective risk assessment strategies, ensuring that debriefings are performed after every day of the air show, or establishing a mechanism by which any safety deficiencies identified during the debriefings are communicated to the FAA and the ICAS to help improve future guidance and policy. (One CAF volunteer stated that debriefings typically did not occur on the last day of an air show. See section 2.4.1 for more information.) The NTSB believes that, although the new guidance materials contain helpful information, as guidance materials—rather than standard operating procedures—they still allow for disparity in how different event organizers or air bosses may choose to manage aircraft separation, and they do not adequately address risk assessment or debriefing strategies to be applied to every performance. Thus, the NTSB concludes that a documented safety risk assessment for each performance would allow air show event organizers and air bosses to identify hazards and determine and apply effective mitigations, such as establishing administrative controls to ensure aircraft deconfliction for performances involving multiple, dissimilar aircraft.

[ANALYSIS] Although CAF’s chief aviation officer stated that the scanners’ role during the air show included looking for traffic, per the CAF’s loadmaster operations manual and training agenda, their primary in-flight responsibilities involved monitoring engine and tailwheel operations, and there were no procedures or training specified for assisting with collision avoidance. Although the toxicology results for one scanner indicated that he had used multiple medications that were potentially impairing or that might indicate potentially impairing underlying medical conditions, given his limited role on the accident flight, it is unlikely that any impairment, if present, could have affected the flight’s outcome.51 Thus, the NTSB concludes that no pilot qualification deficiency or airplane malfunction or failure was identified, and there is no evidence that any flight crewmember’s medical condition or use of medications contributed to the accident. 2.2 Accident Sequence As part of the performance, the Boeing B-17G was in the first position of five historic bomber airplanes flying as multiple solo aircraft in trail, and the Bell P-63F was in the last position of three historic fighter airplanes flying in formation. Generally, the pilots in each group (bombers and fighters) flew multiple passes parallel to runway 13/31 at KRBD and completed repositioning turns on the south side of the extended show lines to set up for each respective pass, as directed by the air boss. The air boss provided directives based on his continuous, real-time assessment of all the variables associated with the performance, including the respective positions of the airplanes. The airplanes’ altitudes and airspeeds varied throughout the performance, which included descending passes from the bomber group, fighter formation maneuvers, the coordinated detonation of ground pyrotechnics, and a 50 For the flight engineer, this included a negative ethanol result in vitreous, which is typically least susceptible to postmortem microbial ethanol production (Kugelberg and Jones 2007, 10–29). 51 Although the scanners’ respective seating locations on board the accident airplane (right or left door position) were not known, given the rapid approach of the Bell P-63F from behind and above the Boeing B-17G, there would likely have been insufficient time for either scanner to observe its approach, recognize it as a threat, and communicate that to the flight crew, and for the flight crew to react. Aviation Investigation Report AIR-24-07 56 real-time narration from an air show announcer on the ground. In addition, the air boss coordinated the flight of CAF-operated Boeing PT-17, which was providing a revenue passenger ride during the performance, and the taxiing of a CAF-operated Boeing B-29 that was positioning for takeoff to join the performance. The air boss, who stood atop a set of air stairs on the field, directed all these activities via radio on the air show radio frequency. Position data were available for seven of the eight airplanes in the performance (all but the bomber in position 5) and the Boeing PT-17 ride flight.

[ANALYSIS] Examination of the horizontal stabilizer indicated that there was no preimpact malfunction or failure of the horizontal stabilizer. Numerous examinations of the wreckage failed to produce evidence of preimpact structural problems. Engine mount separation points showed evidence of impact overload. All doors were closed and latched. The Safety Board considered the possibility that the “mist” trailing the wing observed by witnesses was produced by fuel or hydraulic fluid resulting from a structural failure of some nature. However, the investigation disclosed no evidence of a structural failure that would have allowed fuel or hydraulic fluid to escape. 2.4 Systems From the flightcrew conversations recorded on the CVR and the flightpath described by FDR data, it is evident that the loss of control occurred suddenly and that the Analysis 119 Aircraft Accident Report crew were not aware of any prior problems with the airplane's systems. However, the lateral upset and the flightpath of the airplane during the final 9 seconds of flight could have resulted from a flight control system malfunction. Thus, the Safety Board's investigation focused on an examination of the wreckage and all recovered components of the airplane's hydraulic and flight control systems in an effort to identify any anomalies that could have produced the loss of control. The onset of the loss of control occurred nearly 30 seconds after the flaps were extended to 30 degrees. The trailing edge flaps and leading edge devices would have began extending immediately and would have reached the command position before the first officer's comment, “we're at a thousand feet,” which was made in a tone of voice that did not express unusual alarm. Thus, the Safety Board concludes that the flap operation was symmetrical and normal. 2.4.1 Hydraulic Power The primary flight controls of the B-737 are powered by the independent A and B hydraulic systems previously discussed in section 2.2. A loss of fluid or pressure from either of these systems would result in a loss or degradation of some flight control functions. However, the Safety Board found no indications that the systems had malfunctioned, except for a stretched bulb filament in the HYD indicating light on the first officer's annunciator panel. Because several other light bulb filaments were stretched, some of which would normally illuminate only in a press-to-test check, the Board does not view this evidence as meaningful. The evidence also shows that the motor-operated shutoff valves in both the system A and System B flight control modules were open and that the motor-operated shutoff valves in the standby hydraulic system module were off or closed. Because impact loads do not usually affect the position of motor-operated valves, it is assumed that the systems were operated in this normal configuration before impact. Had the flightcrew been aware of an A or B hydraulic system problem, it would be expected that they would have talked about it and perhaps selected the standby system. Thus, the Safety Board believes that the A and B systems were pressurized and capable of delivering hydraulic power to the flight controls. The teardown examination of the hydraulic components showed considerable evidence of contamination in the A, B, and standby systems.

[ANALYSIS] Had the flightcrew been aware of an A or B hydraulic system problem, it would be expected that they would have talked about it and perhaps selected the standby system. Thus, the Safety Board believes that the A and B systems were pressurized and capable of delivering hydraulic power to the flight controls. The teardown examination of the hydraulic components showed considerable evidence of contamination in the A, B, and standby systems. Most of the contaminants were portions of “O” rings or backup rings that had migrated through the system and were trapped in filter housings. In those cases where contaminants were found to potentially affect the function of relief or check valves, it was determined that there would have been no effect on essential flight control components. While the level of contamination in the hydraulic systems of this airplane seemed excessive, the Safety Board did not determine whether the level was atypical to that which would be found on other airplanes of comparable vintage. Analysis 120 Aircraft Accident Report 2.4.2 Flight Control Systems From the FDR data, it is apparent that the airplane's departure from controlled flight began with a sudden heading change to the right. A lateral or directional flight control problem could produce such a maneuver whereas a longitudinal control system malfunction would produce a pitching maneuver evident by a sudden change in the airplane's load factor. A pitch change was not evident in the FDR data. There were no anomalies found in the longitudinal flight control components that were available for examination. The elevators were recovered at the accident site and the horizontal stabilizer was trimmed in a normal range. During the attempted recovery from the upset, the airplane's load factor increased to about 4 G--a maneuver that would have required a pilot-commanded elevator deflection. The Safety Board thus concludes that the elevator control system was functional until impact. The lateral control system consists of ailerons and flight spoilers controllable by the captain's and first officer's control wheels. The aileron power control units provided evidence that the ailerons were at or near neutral at impact. There were no anomalies noted in the actuators that could account for an uncommanded movement. Although there was some conflicting evidence regarding flight spoiler position, all of the damage was consistent with impact-applied loads. The aileron spring cartridge, which is installed to permit independent operation of the left or right ailerons in the event that the opposite side of the aileron system becomes jammed, was bent and extended. This damage also was readily explainable by impact loading and is not viewed by the Safety Board as evidence of an in-flight problem. Thus, there was no evidence that a lateral control system malfunction occurred in flight. There is also no evidence that a ground spoiler deployed to cause the lateral upset. The condition of the ground spoiler control valve slide was consistent with a retracted spoiler position. Further, had either the flight or ground spoilers been extended in flight, the airplane would not have been able to achieve a 4-G load factor at 212 KIAS without activating the stall warning stick shaker. The sound of the stick shaker was not heard on the CVR.

[ANALYSIS] Since there is no evidence of an interruption of electrical power to either of the recorders, the Safety Board concludes that the electrical busses and the respective electrical motor-driven hydraulic pumps remained powered throughout the flight. With both the A and B hydraulic systems operating, it can be assumed that, absent some other unidentified failure, there was sufficient control capability to cope with any combination of engine thrust variations. Thus, while the Safety Board cannot rule out the possibility of engine surges or a momentary asymmetric thrust condition, the Board concludes that these factors, if they did occur, should not have resulted in the loss of control evident in this accident. Analysis 118 Aircraft Accident Report 2.3 Structures All of the airplane's flight control structure was found and examined, except for a portion of the rudder and vertical stabilizer. The wreckage was localized, and there was no wreckage found along the flightpath. The portion of rudder and vertical stabilizer not examined consisted of composite material located in the middle of the surfaces. Fragments of charred composite fabric were found with the extremities of the surfaces indicating that they were present at impact and burned during the postcrash fire. Reconstruction of the wing structure indicated that all of the parts were attached until impact. Examination of fractures of the wings indicate that the failure modes were consistent with impact overload failures. Examination of the wing flaps showed that the flaps were attached to the wing structure and there were no mechanical failures prior to impact with the terrain. The positions of all of the flap jack screws indicated that there was not a split flap condition and the flaps were at the 10-degree position at impact. This position was further confirmed by metallurgical analysis of the detent track from the flap handle module. Examination of the ailerons indicated that they were attached to the wings until impact. The continuity of the flight control cables throughout the wings indicated that the aileron cables did not malfunction in the wing areas. The attach points for the spoilers indicated that they were attached to the wing structure until impact. Crushing of the leading edge devices (slats) in the extended position indicated that all of the slats were properly extended at impact. Reconstruction of the empennage revealed that all parts were attached to the structure until impact. The recovery of the rudder top cap and the balance weights at the crash site indicated that the rudder was present and intact at impact. The recovery of the elevator end balance weights and the elevator hinges at the crash site indicated that the elevators were present and intact at impact. Examination of the elevator control mechanisms indicated that there was no elevator malfunction prior to impact. Examination of the horizontal stabilizer indicated that there was no preimpact malfunction or failure of the horizontal stabilizer. Numerous examinations of the wreckage failed to produce evidence of preimpact structural problems. Engine mount separation points showed evidence of impact overload. All doors were closed and latched. The Safety Board considered the possibility that the “mist” trailing the wing observed by witnesses was produced by fuel or hydraulic fluid resulting from a structural failure of some nature.

[ANALYSIS] The focus of the 116 Aircraft Accident Report investigation and analysis therefore centers on events that might have produced rolling moments greater than those that can be countered by the B-737's lateral control system. If control countermeasures were applied in a rapid manner, only large sideslip angles, severe atmospheric disturbances, control system anomalies or structural failures could produce rolling moments greater than the restoring capacity of the airplane's lateral control system. In addition, if the crew used rudder control to either reduce a potential sideslip or create a sideslip angle aiding in roll recovery control, then the upsetting event had to be even more severe than that which could be corrected by control wheel alone. The Safety Board attempted to determine an identifiable reason for the loss of control of flight 585 and the inability of the flightcrew to prevent the accident. During the course of the investigation and analysis of the available data, several possible scenarios were considered. These scenarios included loss of directional control (uncommanded rudder deflection); loss of lateral control (failure in the lateral control systems—flaps, slats, spoilers, and ailerons); atmospheric disturbances (windshears or rotors); or a combination of airplane malfunctions, atmospheric disturbances, structural failures, engine failures, or flightcrew performance. 2.2 Engines The Safety Board considered the possibility that one or both of the engines malfunctioned during the final portion of the flight and initiated a loss of control or prevented the flightcrew from maintaining control. This analysis included examination of the evidence from the wreckage, the CVR spectral signatures, and aerodynamic simulation. The postcrash examination of the engines, as well as the indications on the engine pressure ratio (EPR) gauges and transmitters, showed that the engines were developing power at the time of impact. The evidence was conclusive and the indications of power were similar in both engines. Nonetheless, there is some evidence to support a theory that one or both engines had flamed out in flight, caused control difficulties, and then recovered to normal operation before impact. The CVR spectral analyses show two separate signatures consistent with engine characteristic frequencies, just prior to the comment “Oh God.” The frequencies indicate the engines were developing nearly equal thrust at that time. The signatures disappear in the foreground noise at the “Oh God” comment and are not seen for several seconds thereafter. Four or five seconds prior to impact, two signatures were noted that are consistent with two engines accelerating with one engine leading the other by 2 seconds. However, the gaps in the spectral traces preclude firm conclusions that the observed traces were from the engines. Also, some witnesses reported hearing popping or cracking sounds coming from the airplane when it was about 1/2 mile from the crash site. Witnesses also reported observing a “mist” trailing the airplane's right wing. Both the sounds and the mist could Analysis 117 Aircraft Accident Report have been associated with engine surges (compressor stalls) that could have accompanied an attempted relight and acceleration of engines in the presence of turbulent air. However, engine thrust variations alone, even with a total flameout, cannot explain the loss of lateral control.

[ANALYSIS] Although the testing determined that this mechanism could cause a rudder reversal, Boeing indicated that subsequent design changes in the servo valve eliminated this possibility. Analysis 137 Aircraft Accident Report the other is a jam of the primary to the secondary slide of the main rudder PCU servo valve combined with a jam of the secondary slide to the servo valve housing at positions other than neutral (known as a dual jam). These failure mechanisms probably did not play a role in the USAir flight 427, United flight 585, and Eastwind 517 upsets.109 Nonetheless, the failure mechanisms are cause for concern because they further illustrate the vulnerability of the 737 rudder system to jams that could produce rudder deflections and result in catastrophic consequences. 2.8 Flight Data Recorder For updated information about FDR capabilities, see the Safety Board’s final report of the USAir flight 427 accident. 109 The Safety Board considers that a dual slide jam is a less likely accident scenario than a jam of the secondary slide to the servo valve housing because the dual jam would require two extremely rare failures to exist in the servo valve at the same time. 138 Aircraft Accident Report 3. Conclusions Note: Because the Safety Board’s analysis of the United flight 585 accident also included analysis of the USAir flight 427 accident and the Eastwind flight 517 incident, some of the findings pertain to these two events. 3.1 Findings 1. The flightcrew was certificated and qualified for the flight. 2. The airplane was properly certificated and maintained in accordance with existing regulations. Maintenance actions to correct the previous discrepancies related to uncommanded rudder inputs were proper and in accordance with maintenance manual procedures. 3. The airplane was dispatched in accordance with company procedures and Federal regulations. Dispatch of the airplane with an inoperative APU generator was not a factor in the accident. 4. There was no evidence that the performance of the flightcrew was affected by illness or incapacitation, fatigue or problems associated with personal or professional backgrounds. Procedures and callouts were made in accordance with UAL procedures. 5. There were no air traffic control factors in the cause of the accident. 6. There was no evidence of any preimpact failure or malfunction of the structure of the airplane or of the airplane's electrical, instrument, or navigation systems. 7. Both engines were operating and developing power at the time of impact. 8. The crew did not report any malfunction or difficulties. 9. Galling found on the input shaft and bearing from the standby rudder actuator power control unit could not cause sufficient rudder deflection to render the airplane uncontrollable. 10. It is very unlikely that the loss of control in the United flight 585 accident was the result of an encounter with a mountain rotor. 11.

[FINDINGS] It is very unlikely that the loss of control in the United flight 585 accident was the result of an encounter with a mountain rotor. 11. Analysis of the cockpit voice recorder, National Transportation Safety Board computer simulation, and human performance data (including operational factors) from the United Airlines flight 585 accident shows that they are consistent with a rudder reversal most likely caused by a jam of the main rudder power control unit servo valve secondary slide to the servo valve housing offset from its neutral position and overtravel of the primary slide. Conclusions 139 Aircraft Accident Report 12. The flight crew of United flight 585 recognized the initial upset in a timely manner and took immediate action to attempt a recovery but did not successfully regain control of the airplane. 13. The flight crew of United flight 585 could not be expected to have assessed the flight control problem and then devised and executed the appropriate recovery procedure for a rudder reversal under the circumstances of the flight. 14. Training and piloting techniques developed as a result of the USAir flight 427 accident show that it is possible to counteract an uncommanded deflection of the rudder in most regions of the flight envelope; such training was not yet developed and available to the flight crews of USAir flight 427 or United flight 585. 15. It is possible that, in the main rudder power control units from the USAir flight 427, United flight 585, and Eastwind flight 517 airplanes (as a result of some combination of tight clearances within the servo valve, thermal effects, particulate matter in the hydraulic fluid, or other unknown factors), the servo valve secondary slide could jam to the servo valve housing at a position offset from its neutral position without leaving any obvious physical evidence and that, combined with a rudder pedal input, could have caused the rudder to move opposite to the direction commanded by a rudder pedal input. 16. The upsets of USAir flight 427, United flight 585, and Eastwind flight 517 were most likely caused by the movement of the rudder surfaces to their blowdown limits in a direction opposite to that commanded by the pilots. The rudder surfaces most likely moved as a result of jams of the secondary slides to the servo valve housings offset from their neutral position and overtravel of the primary slides. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of the United Airlines flight 585 accident was a loss of control of the airplane resulting from the movement of the rudder surface to its blowdown limit. The rudder surface most likely deflected in a direction opposite to that commanded by the pilots as a result of a jam of the main rudder power control unit servo valve secondary slide to the servo valve housing offset from its neutral position and overtravel of the primary slide. 140 Aircraft Accident Report

[ANALYSIS] The Safety Board concludes that analysis of the CVR, Safety Board computer simulation, and human performance data (including operational factors) from the United flight 585 accident shows that they are consistent with a rudder reversal most likely caused by a jam of the main rudder PCU servo valve secondary slide to the servo valve housing offset from its neutral position and overtravel of the primary slide. Also, because the United flight 585 upset occurred when the airplane was less than 1,000 feet above the ground, the pilots had very little time to react to or recover from the event. Thus, the Safety Board concludes that the flight crew of United flight 585 recognized the initial upset in a timely manner and took immediate action to attempt a recovery but did not successfully regain control of the airplane. The Safety Board further concludes that the flight crew of United flight 585 could not be expected to have assessed the flight control problem and then devised and executed the appropriate recovery procedure for a rudder reversal under the circumstances of the flight. The Safety Board also concludes that the training and pilot techniques developed as a result of the USAir flight 427 accident show that it is possible to counteract an uncommanded deflection of the rudder in most regions Analysis 135 Aircraft Accident Report of the flight envelope; such training was not yet developed and available to the flight crews of United flight 585 and USAir flight 427.103 2.7 Rudder System Jam Scenarios During the investigation of the USAir flight 427 accident, the Safety Board examined the rudder systems of the USAir flight 427, United flight 585, and Eastwind flight 517 airplanes and was unable to identify any obvious physical evidence that a jam occurred within the servo valves. Further, the investigation had not revealed how the secondary slide could jam to the servo valve housing under conditions that would normally be encountered by an airplane in air carrier operations and not leave any physical evidence that the jam occurred. However, the Safety Board demonstrated that, in servo valves with tight clearances,104 the secondary slide could jam to the servo valve housing and leave no physical evidence of that jam (albeit under thermal conditions that would not normally be encountered by an airplane in air carrier operations). Further, small particulate matter in the hydraulic fluid could reduce the already tight clearances in the servo valve, requiring less of a thermal differential for the valve to jam. In addition, it is possible for a large amount of small particles to provide the jamming potential of a larger, stronger piece of metal without leaving a mark.105 Further, testing showed that, when the secondary slide was jammed to the servo valve housing and a sufficiently high-rate force was applied on the input crank, compliance within the rudder system could allow the primary slide to overtravel and result in a reverse rudder command.

[ANALYSIS] Witnesses also reported observing a “mist” trailing the airplane's right wing. Both the sounds and the mist could Analysis 117 Aircraft Accident Report have been associated with engine surges (compressor stalls) that could have accompanied an attempted relight and acceleration of engines in the presence of turbulent air. However, engine thrust variations alone, even with a total flameout, cannot explain the loss of lateral control. Simulator tests showed that the asymmetrical thrust differences produced by a failure of one engine or a 5-second split in engine acceleration were easily handled with flight controls assuming all hydraulics systems were operational. The simulator tests showed that thrust differentials consistent with the signatures from the CVR would produce some yawing and rolling moments. However, at the airspeeds recorded on the FDR, the effects of asymmetric thrust would have been minimal and well within the capability of the airplane's lateral and directional control systems. The Safety Board also considered the effects that a failure of one or both engines would have on the airplane's hydraulic systems. In the B-737, the A hydraulic system is powered by engine-driven hydraulic pumps on both engines. Either pump is capable of maintaining the system operating pressure while delivering 22 gallons per minute flow. At engine windmilling speed, the flow capability of the pumps drops to 4 or 5 gallons per minute. However, even with maximum utilization of the flight controls, including a simultaneous flap retraction, the flow requirement would be about 4 gallons per minute. Thus, even with a complete flameout of both engines, there should be adequate hydraulic power available to the A hydraulic system to provide for flight control. Also, as an engine accelerates from windmilling speed, the flow capacity of the engine-driven pump increases immediately. Further evidence of normal hydraulic capability on the A system was provided by the elapsed time for flap retraction. The time from the sound of flap handle movement recorded on the CVR to impact was consistent with the normal flap retraction speed from the 30-degree position to the 10-degree position as found after the crash. The B hydraulic system is powered by two electrical motor-driven hydraulic pumps, each of which is capable of maintaining system pressure while producing 6 gallons per minute flow. It is possible that, at engine windmilling speed, the constant speed drive would not maintain the electrical generator frequency and the associated electrical buss would drop off line. However, if this had occurred, the FDR and CVR, which are powered by the same busses, would have ceased to operate. Since there is no evidence of an interruption of electrical power to either of the recorders, the Safety Board concludes that the electrical busses and the respective electrical motor-driven hydraulic pumps remained powered throughout the flight. With both the A and B hydraulic systems operating, it can be assumed that, absent some other unidentified failure, there was sufficient control capability to cope with any combination of engine thrust variations.

[ANALYSIS] There is also no evidence that a ground spoiler deployed to cause the lateral upset. The condition of the ground spoiler control valve slide was consistent with a retracted spoiler position. Further, had either the flight or ground spoilers been extended in flight, the airplane would not have been able to achieve a 4-G load factor at 212 KIAS without activating the stall warning stick shaker. The sound of the stick shaker was not heard on the CVR. The simulation conducted during the investigation determined that a 20-degree or greater deflection of the rudder to the right could induce extreme control difficulties and could lead to a rolling moment consistent with that observed by witnesses and determined during flightpath analysis of this accident. The Safety Board was therefore concerned about the previous maintenance discrepancies relating to rudder operation on the accident airplane. Previous discrepancies: The first evidence of a potential rudder control problem on N999UA occurred on February 25, six days before the accident flight, when the flightcrew on that day experienced a transient uncommanded yaw to the right. The crew turned off the yaw damper and no further uncommanded yaws were observed during the flight. Following that flight, UAL maintenance replaced the yaw damper coupler. However, on February 27, another crew experienced an uncommanded yaw to the right, Analysis 121 Aircraft Accident Report and they, too, turned off the yaw damper to eliminate a recurrence of the problem. The UAL maintenance personnel then replaced the yaw damper transfer valve in the rudder MPCU. No further problems were encountered prior to the accident flight. The Safety Board believes that the UAL maintenance efforts to troubleshoot the system were in accord with normal practices. However, it is doubtful that these actions corrected the problem since subsequent tests of both of the removed components showed that they operated normally. During the examination of the MPCU recovered from the wreckage, it was noted that one of the electrical wires to the solenoid was loose and circuit continuity was intermittent. The Safety Board believes that this intermittent circuit could have been the cause of the uncommanded yaws experienced on the earlier flights. If this were the case, the effect of the discrepancy would be erratic deflections of the rudder when the yaw damper was in use. However, by design, the authority of the yaw damper is limited to 2 degrees of rudder travel. While uncommanded rudder movements of 2 degrees or less could produce noticeable side loads, they would have little or no effect on airplane controllability. Standby rudder actuator input crank shaft galling: The Safety Board believes that the binding of the input shaft to the bearing that is threaded in the standby actuator body could also have produced the two transient uncommanded yaws experienced during previous flights. As discussed in section 1.16.3, a rudder movement initiated by the yaw damper will produce a small angular movement of the standby actuator input crank. If the crank is not free to move relative to the actuator body, the feedback loop to the MPCU servo valve will be affected so that a rudder deflection command signal may be applied to the MPCU through rotation of the torque tube.

[ANALYSIS] Further, Safety Board testing showed that the main rudder PCU servo valve could jam, without leaving a physical mark, in a way that could lead to rudder reversal. Additionally, in all three upset events, the available human performance data comported well with a rudder system malfunction but were inconsistent with an inappropriate pilot input (or a rotor in the case of United flight 585). The statements of the Eastwind flight 517 flight crew were fully consistent with an uncommanded rudder input. In addition, the Safety Board’s and Boeing’s computer simulation and kinematic studies both indicated that, in the Eastwind flight 517 incident, the rudder moved to a position consistent with rudder reversal but inconsistent with a normally operating rudder system (given the pilots’ consistent recollections of the captain applying great force to the rudder pedals). Therefore, the Safety Board concludes that the upsets of USAir flight 427, United flight 585, and Eastwind flight 517 were most likely caused by the movement of the rudder surfaces to their blowdown limits in a direction opposite to that commanded by the pilots. The rudder surfaces most likely moved as a result of jams of the secondary slides to the servo valve housings offset from their neutral position and overtravel of the primary slides. In addition to this reversal potential, the Safety Board’s investigation revealed two other potential failure mechanisms108 within the 737 rudder control system that could result in a deflection to the rudder’s blowdown limit. One of these potential failure mechanisms is a physical jam in the rudder system input linkage (between the PCU’s input crank and body stop), preventing the main rudder PCU control valve from closing; 107 The aerodynamic blowdown limit for the Eastwind flight 517 incident airplane, assuming normal (unreversed) operation, far exceeded the 6 to 6.5° rudder deflection that was apparently involved in the incident. According to Boeing, the rudder blowdown limits for the incident airplane (including the reduced hinge moment from the PCU’s excessive leakage at the bypass valve) would have been about 9° when operating at 250 knots under normal (unreversed) pilot command, with the variation depending on the sideslip value. Given the airplane’s leaking bypass valve and the reduced rudder hinge moment from a secondary slide jam at 55 percent from neutral), the aerodynamic blowdown limit would have been about 6.5°. 108 A third potential failure mechanism—a jam of the primary to the secondary slide with overtravel of the secondary slide—was identified as a result of testing after the July 1992 United Airlines rudder anomaly that occurred during a ground check. Although the testing determined that this mechanism could cause a rudder reversal, Boeing indicated that subsequent design changes in the servo valve eliminated this possibility.

[CONCLUSIONS] The flight appeared to have been normal up to an altitude of about 600 feet a.g.1. and near the end of the runway when the captain routinely acknowledged an instruction ta contact departure control, The first Separation oceu.:.4 about 6,000 feet beyond the end of the runway and about 25 to 30 seconds after the time the airplane pessed the end of the runway. The accident was considered to be nonsurvivable because the impact forces exceeded the limitations of human tolerance and the decreased cabin volume was insufficient to support human life. The flight crewmembers were properly certificated, No medical or psychological conditions were found which might have adversely affected the flightcrew's performance. Both engines were operating normally untii impact. The propellers were intaet and undamaged until impact. All fractures and deformations of the right engine mounts resulted from impant. The engine mounts were hot subjected to any centrifugally induced vibration forces, There was no evidence of any turbulence or windshear at the time of the accident, There wus no evidence of any significant turbulence on the previous flight There was no damage to the structure which might suggest a pre-flight collision with another vehicle. The stabilizer forward attachment structure was fully capable of carrying its ultimate design loads. ae or oe a mate le i al 16, 17, 18, 19, 20. 21. 22. 23. 24. 25, 26. 27. 28. ~44- utter. The separation of the horizontal stabilizer was caused by an abnormal asym metrical afr load on the stabilizer. Structural failure was preceded by some other failure or malfunction of the airplane's elevator control system. The aft fracture of the left elevator control rod was due to compression buekling at or near the midpoint. The aft fracture of the left elevator control rud occurred before the elevator separated from the stabilizer. The elevator control rod failed in compression buckling with an applied load of about 466 pounds. The elevator control rod would fracture at its midpoint when the load is applied through the rod end bearings, as ii would be applied in the norma] ‘light through pilot input. A combination of commanded aircraft hose up pitch attitude and nose down trim tab deflection results in compression forces in both the left and the right control rods, with the force in the left rod being the greatest. The control column pull force required to cause a compression fallure of the left elevator control rod would approach or exceed the maximum two-hand pull force of about 20% pounds, which can be applied by one male pilot of average strength. The trim tab actuator Indicated full trailing edge up trim tab deflection (airplane nose down). The separation of the elevator tips from the elevators and the elevators from the stabilizer occurred during or immediately after the horizontal Stabilizer attachment failed and as a result of inertial and aerodynamic loads which were imposed on the stabilizer and elevator assembly during its separation from the fuselage. The left elevator trim tab requires about 30 seconds to travel from an approximately neutral takeoff trim position to the full trailing edge up position.

[ANALYSIS > THE ACCIDENT] Moreover, the manufacturer's tests in which the horizontal stabilizer and elevator structures v..~¢ instrumented to measure the vibration loads caused by propeller imbalance disclosed that high loads sufficient to damage the elevators, could be produced only with a 14-inch or longer length of one propeller blade missing. However, the manufacturer stated that a propeller imbalance of this magnitude also would cause destruction of the engine mounting structure. All fractures and deformations of the engine attachment structure on N96PB were typical of damage produced by the extreme forces generated during ground impact and net those that would have been generated by a damaged propeller. Consequently, the Safety Board concludes that there were no destructive vibratory loads imposed on the horizontal stabilizer structure attributable to propeller imbalance. 2.9 Abnormal Stabilizer Loading Caused by Flight Control Malfunctions A significant investigetive finding resulted from the tests conducted by the manufacturer when abnormal asymmetrical loads were statically applied to the horizontal stabilizer. Upon application of loads approximating those air loads produced with full antisymmetric elevator deflections at 144 knots (or with lesser elevator deflections at higher speed), the stabilizer forward attachment fitting and the bulkhead No. 33 structure on the test fixture deformed and fractured in a manner nearly identical to the deformation and fractures evident on N96PH. The test provided strong evidence that the a I 2 separation of the horizontal stabilizer of N96PB at the stabilizer forward attachment fitting could have been caused by an abnormal asymmetrical air load on the stabilizer. An asymmetrical sir load of this magnitude will occur only with antisymmetric elevator deflection, a circumstance which can only follow some other failure or malfunction of the airplane's elevator control system. Therefore, the Bourd believes the test showed that a failure of the contrat system praceded the structural soparation of the stabilizer, 2.19 Left Control! Rod Fracture Although the left and right elevator actuating arms were not interconnected on N96PB, the two elevators are connected by the aft belleranks which transmit control system motion to the forward end of the left and right elevator control rods. Consequently, differential deflection of the left and right elevators requires failure of an aft bellcrank, an elevator control rod, cr an elevator actuating arm. There was no evidence of failure of either of the aft belicranks or either of the elevator actuating arms. However, both elevator control rods were fractured, Based on its examination of the fractures and the relative position of adjacent fuselage structure, the Safety Board concludes that the symmetrically located fractures of the left and right elevator control rods (11 inches aft of the aft bellerank attachments) occurred when a channel section at fuselage bulkhead No. 34 sliced the rods as the leading edge of the horizontal stabilizer moved downward and aft during its separation from the fuselage. There were no other fractures in the right contro] rod indicating that the rod was intact until the stabilizer separated.

[ANALYSIS > THE ACCIDENT] However, the manufacturer indicated to the Safety Board that the weight distribution of the accident airplane would not have a significant effect on its elevator structure. ) seer , sc Sek SS Sa Ads oot en ags oan, rcnee One aye A I — peas to nme aa! The consultant's analysis of flutter characteristics of the empennage considered a reduction in stiffness attributable to loose rivets at bulkhead No. 3., a completely separated elevator hinge, the actual balance weight distribution of N96PB, and a broken elevator control rod. The analysis showed that the airplane would not have encountered an aerodynamic flutter condition in the speed range between 50 and 200 knots. Therefore, the Safety Bow'd concludes that the structural failure was not caused by a divergent aerodynamie flutter. Although analys?s and tests showed that the existence of loose or sheared fastenets in the bulkhead No. 33 structure of W96PB did not affect the ability of the structure to withstand applied static loads or the airplane's aeroelastic characteristics, the Safety Board remains concer-ed that this condition on other EMB-110P1 and P2 airplanes could lead to progressive fatigue and premature structure failure. The Safety Beard believes that the FAA should require the horizontal stabilizer attachment structure of EMB-110P1 and P2 airplanes be modified to preclude such damage in accordance with a procedure set forth by the manufacturer. The Safety Board agrees that the tests showed that the stabilizer forward attachment structure at bulkhead No. 33 would carry ultimate stabilizer loads even though weakened by cracks and the removal of fasteners in the bulkhead web. Nevertheless, the Safety Board is concerned that the tests were not sufficient tu show conclusively whether the resulting change in load distribution would affect the fatigue life of the redundatit load path. 2.8 Vibratory Load Considerations The missing part of one blade of the right propeller prompted concern that the blade might have been damaged before or during takeoff and that 4 resultant imbalance might have caused structural failure of the horizontal stabilizer. The damage to the other blades of the right propeller and to the blades of the left propeller were not typical of damage which would be expected from a takeoff ground strike. Further, the end of that portion of the blade on the right propeller which remained attached to the hub had melted and sagged under its own weight during exposure to the posterash fire. Therefore, the Safety Board believes that the missing portion of the blade was consumed in the ground fire. Moreover, the manufacturer's tests in which the horizontal stabilizer and elevator structures v..~¢ instrumented to measure the vibration loads caused by propeller imbalance disclosed that high loads sufficient to damage the elevators, could be produced only with a 14-inch or longer length of one propeller blade missing. However, the manufacturer stated that a propeller imbalance of this magnitude also would cause destruction of the engine mounting structure.

[ANALYSIS > THE ACCIDENT] Turbulence near the hae ae! ~Q.. ground was probably no greater than moderate. Further, there was m. evidence that the q airplane had encountered significant turbulence during the previous flight. Therefore, the Z Safety Board concludes that weather was not a factor in this accident. 4 2.6 Preexisting Condition of Airplane Structure The investigation determined that before takeoff there was no Gamage t¢ either the stabilizer or to the elevator components sufficient to suggest a preflight a collision between the airplane and another vehicle, such as a fuel truck or baggage cart. . Further, those persons who observed and serviced the airplane during its Jacksonville turn * around did not see any vehicle come in contact with the airplane. No baggage cart was Euse, Therefore, the Safety Board concludes that the airplane was not subjected ta external loads of sufficient magnitude to produce a deformation or failure of the stabilizer attachment structure during « preflight collision, such as contact with a ground vehicle. The Safety Board determined that the fractures of the male lugs of the i forward attachment fitting at buikhead No. 33 eaused the horizontal stabilizer assembly a to separate from the fuselage. The deformation and fractures of the attachment fitting a indicate that the male lugs fractured first in shear and then in tensile overstress as the c horizontal stabilizer moved aft and twisted clockwise (looking forward) relating to the Sy fuselage, Because the loads on the forward attachment fitting are carried into the a fuselage monocoque structure by rivets and channels at bulkhead No. 33, and because = there was evidence of preexisti damage in (sis area, the effects of such preexisting Ei damage on the load carrving capability of this stcueture were analyzed in depth. Z The fretting around some of the fastener holes in the channels which transmit ia the loads from the upper right corner of the forward attachment fitting into the fuselage forward of bulkhead No. 33 indicated that some fasteners had been loose before the fina) structural failure. Looseness in these attachments would have resulted in a transfer of inereasad stabilizer loads into the bulkhead No. 33 web. The small pre-existing fatigue erack ($/16-ineh long) in the bulkhead web supported the contention that loose or sheared 4a rivets had bewn present before the accident and that the web hud »een exposed to excess E stress, I I The susceptibility of the EMB-110P1 and P2 models to fastener distress and web fatigue cracking in the bulkhead No. 33 structure was known before the accident. ‘” The knowledge had prompted the manufacturer to issue a service bulletin which described f an inspection program to deteet loose fasteners and web fatigue cracks. The service bulletin also described an alternative modification to correct the loose fasteners and a procedure t. repair the web cracks.

[ANALYSIS > THE ACCIDENT] There were no other fractures in the right contro] rod indicating that the rod was intact until the stabilizer separated. A similar conclusion regarding the left control “0d could not be made because that control rod was fractured in two places with a 9-inch intervening section missing. The aft failure of the left control rod was typical of compression buckling and was initially attributed to impact forces applied when the rod struck the ground and was foreed into the earth, However, after determining that differential elevator deflection could explain the horizo ward attachment separation, the left control rod fracture and the fuse : examined more closely for evidence that the left elevator control rod fractured during flight. The examination disclosed that two facts Supported an In-flight fracture: (1) the compression buckling fracture occurred at or very near to the exact midpoint of the control rod (a failure which would be typical of a control system compression induced fracture); and (2) there was an impact mark on the aft side of an upper channel at fuselage bulkhead No. 35 which matehed the shape of the fracture surface of the aft position of the left control rod. This impact mark indicated that the rod fractured and the aft portion of the rod had struck the chunnel before tie elevator separated from the stabilizer. Consequentiy, the Safety Board concludes that the left elevator control rod failed as a result of compression overstress during flight; that this failure, in conjunction with abnormal trim tab deflection, permitted clifferential deflection of the left and right elevators; and that the resultant asymmetrical loads caused the horizontal stabilizer separation. 2.11 Control System Overload A load is applied to the elevator control rods whenever a pilot applies a force to either of the control columns to maneuver the airplane in pitch. The load applied under normal conditions is reacted to by the aerodynamic loads on the elevators which are dependent upon the elevator deflection, elevator trim tab position, aud airspeed, s) ‘ * , : * i a <a 4 a r, 3 i ] ‘ Bt 5 “a i 3 i 3 1 i 3g & +8 ] ca * Lt Ee Re aes Bi 4 i Ret: Pe . wes A Fi Pa if i ~33- During steady state flight, the position of the left elevator trim tab is adjusted to produce an aerodynamic load on the left elevator which balances the aerodynamic load on the right elevator.

[PROBABLE CAUSE] I oe The elevator trim system conformed to certification criteria. The installation of an FDR and CVR would have provided significant clues regarding the cause of this accident and remedied action needed to prevent recurrence. . 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was a malfunction of either the elevator control system or the elevator trim system, which resulted in an airplane pitch control problem. The reaction of the flightcrew to correct the pitch control problem overstressed the left elevator control rod, which resulted in asymmetrical elevator deflection and overstress failure of the horizontal stabilizer attachment structure. The Safety Board was not able to determine the precise problem with the pitch control system. 4. RECOMMENDATIONS On January %, 1985, the National Transportation Safety Board recommended that the FAA: Issue an airworthiness directive (AD) to require that before further commercial operation in the United States, the horizontal stabilizer attachment of EMB-110P1 and -110P2 model airplanes not previously modified in accordance with AD 83-14-09, Amendment 38-4527, paragraph (d) or (e), be Inspected using an improved inspection procedure to enhance detection of loose or sheared rivets, particularly where bulkhead 33 transmits the loads from the stabilizer forward attachment to the fuselage monocoque structure. The inspection procedure shouic! require removal of ~48- controls as needed for access to riveted joints and application of external loads to detect relative movement between structural meinbers. The AD should require that deficiencies detected during inspection be reported to the FAA and that they be corrected in (reese with an approved procedure before further flight. ~-85-( 1) Revise airworthiness directive (AD) 83-14-09 to require within a specified period that the horizontal stabiliz of EMB-110P1 and -110P2 nove}} airplanes be modified in a manner similar to that deseribed in Amendment 39-4527, paragraph (d) or (e), which requires the repair of any cracks in the web of bulkhead 33 and the replacement of the original "C" ghannels with redesigned channels and modified rivet patterns.

[ANALYSIS > THE ACCIDENT] Further, it is not Ukely that the I pilots could have taken actions to prevent the accident if the control system had seized or jammed, The inability of the Safety Board to determine conclusively the initial event which resulted in the full trailing edge up deflection of the elevator trim tab precluded the Board from citing either runway trim or a jammed control as causal. Consequently, identification of factors which could have been significant to the accident cause or contributing cause was not possible. For example, if the initial ever: was an electrical trim runaway, the Safety Board would focus greater attention on flighterew performance and operator pilot training; and if the initial event was a seized or janimed control system, the accident may have occurred with flawless pilot performance. In the latter ease, the Safety Board would focus more attention on the airplane design, the operator's maintenance and inspection program, and/or the FAA's surveillance of those: programs, even though the Safety Board's investigation did not find significant tangible evidence of deficiencies in any of the areas. 2.18 Review of FAA Certification An in-flight structural failure of any airplane in the absence of circumstances to explain an obvious overstress condition always prompts concern about the airplane's original design and certification criteria. In this case, the particular areas of interest rulude the design load criteria, aerodynamic flutter characteristics, elevator control system strength, and elevator trim system runaway protection. The Safety Board reviewed the certification procedures and concluded that the FAA's original U.3. certification of the EMB-110 was procedurally proper and in accordance with the provision for the certification of a product that is manufactured in a foreign country. The Special Certification Review initiated by the FAA and the CTA following the accident provided further assurance that the original certification of the airplane was accomplished in accordance with applicable regulations. ee ae Tae PELLET, SRNR ae Fe ee, ~39-- The Special Certifivation Review team evaluated the design load criteria and the aerodynamic flutter characteristics of the airplane and found only minor diserepancies in the analytical and test data used Initially to show compliance with the FARs. The Safety Boarc concludes that the discrepancies were not relevant to the cause of this accident. Neither the design criteria nor the certification requirements included a structural design load consideration for antisymmetric aerodynamic loading of the horizontal, stabilizer. The Safety Board agrees that because it is not possible to achieve such a ioading condition absent other failures which cou’. render the airplane uncontro‘lable, an antisymmetric loading condition is not a reasonable design consideration. In their evaluation of flutter characteristics, the special certification review team noted that the airplane, although in compliance with the U.S. certification basis specified in the appropriate section of the FAR's effective in September 1989 and during origins! certification, was not in compliance with a recent amendment to the FAR which requires that the airplane be shown to be free from flutter following the failure of a trim tab actuating rod.

[ANALYSIS > THE ACCIDENT] Therefore, the Safety Board concludes that both pilots were pulling on their respective control columns when the left elevator control rod failed. 2.12 levator Trim The mechanical damage to the elevator trim tab actuator rod and the molten metal fused position of the trim system threaded cable on the cockpit pedestal trim wheel were both consistent with a full trailing edge up deflection of the elevator trim tab. The design of the mechanism is such that the position of the cable would not have changed during che structural separation of the horizontal stabilizer or during the subsequent impact unless commanded by one of the pilots. Because of the difficult - ontrel situation which must have existed during and after stabilizer separation, it is improbable that either pilot commanded a trim change. Therefore, the Safety Board concludes that the airplane's ejevator trita tab was fully deflected in the trailing edge up position before the structural failure of the stabilizer occurred. Further, the Safety Board concludes that this trim tab position was a key factor in the sequence of events of the accident. tl ¥ Om Rha mem ed Oar eas LAGOA TN oN avi SALT KR LALAN PAG ROS ERRORS TS UB TY LURE LEANER CBRNE A ee sae Pag 5 MI YE HUT ASEH Zo pO SPOR WL Tae Be MRO IEA Lae nae tale nt er ttirt SAREE AH PARE BI LG ING Bet UIE i SPEER IRE SS LAL CRDEREG ESET LGE NEE GE RIGOR AMOR IRE RR NT ORE toa 5 ~34- 2.13 cf Events Leading to Stabilizer Separation The evidence that the elevator trim tab was deflected to its full trailing edge up position (airplane nose down trim), the left elevator control rod was fractured from compression loading during flight, and the horizontal stabilizer structural attachments were overstressed and separated by asymmetrie aerodynamic air loads is all consistent with and supportive of a definitive failure sequence. The aerodynamie¢ loads on the left elevator as the elevator trim tab deflected upward required reactive forces in the control system to prevent the airplane from pitching nose down. The Safety Board cannot assess — the extent to which pilot forces on the control column, or other forces acting in the. elevator flight control system, prevented the sirplane from pitching down as the trim tab was initially deflected upward. However, one explanation of the observed damage is that, at some instant during the posttakeoff climb, the airplane pitched suddenly nose dcwn and that both pilots reacted to . , upt end high pull forees on their respective control columns, }} I & compression load in the left elevator control rod which exceeded the design strength of the rod and caused it to fracture.

[ANALYSIS > THE ACCIDENT] A pilot's normal and reflexive action to an abrupt nose down pitch change at low altitude would be to rapidly reverse the control column forces and pull back 3 on the control yoke. Consequently, it {{s possible that the sudden pull forces exerted by 4 the pilot(s) would have been sufficient to have failed the left elevator control rod. It is : B: also possible that the pilots' pull forces might have caused the control system to seize or jam again, so that the available pitching moment was limited to the extent that the airplane's descending flightpath could not be corrected, The pilots might then have pulled back on the control yoke to their maximum capability in an attempt to prevent impact with the ground. If the seizure or jam was again relieved, destructive dynamic forces would have been imposed on the left elevator control rod. TANASE ORRTESUR EET ER 4 There was no tangible evidence from the examination of the wreckage of a seized or jammed control system, nor have there been any known occurrences of such problems in the EMB~-110P1 and P2 sevice history. However, there have been some SDRs in which stainless steel elevator cables have become worn where they pass through fairlead blocks near the midsection of the forward-to-aft cable run, an event which can lead to seizure of a cable within a fairlead block. The identification of this problem prompted the manufacturer to issue a service bulletin that recommended the replacement of stainless stee] elevator control cables with harder carbon steel cables, which are more resistent to wear. The aft cables were replaced on N96PB in February 1983, and the cables were inspected In Jctober 1984, with no defects noted. Notwithstanding these maintenance actions, it is possible that a worn cable seized within a fairiead block during the takeoff rotation, particularly since problems have been reported with the carbon steel cables. Also, & control system jarn could have been caused by a foreign object interfering with a cable pulley or by a control coluiun, or by a seized right elevator hinge bearing. Any of these conditions could have resulted in an elevator control system seizure or jam which could have been relieved only by high control system forces or by a momentary reversal of the force applied to the control yoke. In summary, the Safety Board believes that a control system seizure or jam, followed by the foregoing sequence of events would expiain this accident and probably is more easily understood that a runaway trim occurrence, because the pilots should have been able to control a runaway trim by applying the required pull force on the control wheel to prevent loss of control even though they might not have been able to immediately diagnose the nature of the emergency. Further, it is not Ukely that the I pilots could have taken actions to prevent the accident if the control system had seized or jammed, The inability of the Safety Board to determine conclusively the initial event which resulted in the full trailing edge up deflection of the elevator trim tab precluded the Board from citing either runway trim or a jammed control as causal.

[ANALYSIS > THE ACCIDENT] However, one explanation of the observed damage is that, at some instant during the posttakeoff climb, the airplane pitched suddenly nose dcwn and that both pilots reacted to . , upt end high pull forees on their respective control columns, }} I & compression load in the left elevator control rod which exceeded the design strength of the rod and caused it to fracture. With the restraint of the left control rod removed, the left elevator instantaneously reacted to the aerodynamic load produced by the fully deflected trim tab and moved rapidly trailing edge down. Simultaneously, the fracture of the left control rod caused the high puil forces on the pilot control column to transfer fully to the intact right elevator contro] rod, which rapidly forced the right elevator to move trailing edge up. The combination of airspeed, which could have reached at least 170 knots during the initial airplane pitch down maneuver, and differential elevator deflection produced ‘high asymmetrical aerodynamic loads on the horizontal Stabilizer, which exceeded the ‘Strength of the Stabilizer forward attachment structure the horizontal stabilizer separated from the airplane in a ele The Safety Board QV an elevators separat . ately after the horizontal stabilizer attachment separated because of the high inertial and aerodynamic loads imposed on the stabilizer assembly in the Separation process. . Although a logical sequence of failure following the deflection of the elevator trim tab has been established, the event that caused the elevator trim tab on N96PB to deflect to its full trailing edge up position could not be conclusively established. However, the Safety Board considered the possible explanations for events which may have caused the trim tab deflection and narrowed the possibilities to two: malfunction in the trim System itself, which may have caused a runaway malfunction in the airplane's primary elevator control system, which may have prompted the pilot to intentionally command full airplane nose down trim. : 2.14 Runaway Trim Theory The electrical switches on the captain's and first officer's control wheels and I ted electrical wiring for the elevator trim tab were destroyed by the posterash fire. Also, the circuitry in the trim adapter box was damaged. Therefore, the pre-crash condition of these components and their possible effect on the functions of the elcatrion: . trim system could not be determined. . The service history of the EMB~110P1 and P2 trim occurrences of an electrical trim runaway there have been multiple occurrences wher have stuck (failed to return to neutral) following a trim application. In the known cases of SEN IRE NRA EEN LEME AMID KB REE Yt Ap ns pele wan ~ . - BAB SoS A IBGE RISERS CIN TRL MRR ALY LES Se TOE BREE ARL BI, Me a eee ~35- a stuck trim switch, the pilot was able to move the switch back to neutral with his thumb. Consequently, the elevator trim system in the model has not posed any significant problems to the airplane or its pilots.

[CONCLUSIONS > FINDINGS] 3.1 Findings 1. The two-member flightcrew and three flight attendants were trained and qualified to conduct the flight in accordance with Federal regulations. There was no evidence of any medical condition that might have affected the flightcrew’s performance. 2. The air traffic control request to maintain 190 knots to the outer marker did not contribute to the accident because it did not affect crew actions, decisionmaking, or situational awareness. 3. The airplane was certificated and equipped and maintained in accordance with Federal regulations and approved procedures. There is no evidence that mechanical malfunctions or failures of the airplane structures, flight control systems, or powerplants contributed to the accident. 4. Because the captain omitted the “Hydraulics” item on the in-range checklist and the first officer failed to detect the error, hydraulic pressure was not available to lower the landing gear and deploy the flaps. 5. The “Hydraulics” item is placed too low on the in-range checklist, rendering it vulnerable to omission. 6. The captain’s distraction from his duties as pilot-in-command and his disregard for the sterile cockpit rule contributed to the pilots’ failure to detect their hydraulic system configuration error when they selected 5o of flaps. 7. Both the captain and the first officer recognized that the flaps had not extended after the flaps were selected to 15o. 8. The pilots’ lack of previous exposure, either through training or during line operations, to the consequences of improper hydraulic system configuration contributed to their failure to detect their hydraulic system configuration error. 9. The pilots failed to perform the landing checklist and to detect the numerous cues alerting them to the status of the landing gear because of their focus on coping with the flap extension problem and the high level of workload as a result of the rapid sequence of events in the final minute of the flight. 10. Had the landing checklist been properly performed, the flightcrew would have detected the failure of the landing gear to extend. 56 11. Although the first officer was unwilling to overtly challenge the captain’s decision to continue the approach, he did attempt to communicate his concern about the excessive speed of the approach to the captain. 12. There was no compelling reason for the captain’s decision to land the airplane; multiple signals and guidance indicated that the approach should be discontinued, as did Continental Airlines’ standard operating procedures. 13. The flightcrew’s degraded performance is consistent with the effects of fatigue, but there is insufficient information to determine the extent to which it contributed to the accident. 14. There were deficiencies in Continental Airlines’ (COA) oversight of its pilots and the principal operations inspector’s oversight of COA.

[ANALYSIS] The airplane was certificated and equipped and maintained in accordance with Federal regulations and approved procedures. There is no evidence that mechanical malfunctions or failures of the airplane structures, flight control systems, or powerplants contributed to the accident. The evidence indicates that the airplane’s hydraulic system was not configured for landing. Because the hydraulic system remained in the low pressure mode, hydraulic pressure was not available to lower the landing gear and deploy the flaps. The flightcrew failed to detect this configuration error and continued its approach into Houston. Comments on the CVR and postaccident statements by the flightcrew indicate that both pilots recognized that the flaps did not deploy after the flaps were selected to 15o, but the flightcrew did not determine the cause of this problem or execute a go-around. The landing checklist was not performed, and the flightcrew did not confirm that the gear was down and locked. The gear warning horn sounded during the approach, indicating that the landing gear was not extended, but it was ignored. When the airplane descended through 500 feet AFE, it was traveling 84 knots faster than the target airspeed of 132 knots. Although, under COA standard operating procedures, this excessive airspeed mandated that the approach be discontinued, the captain rejected a go-around request from the first officer, who was the flying pilot. The GPWS sounded an alert 19 seconds before impact and was ignored. Unaware that the gear was not down, the captain assumed control of the airplane and made a wheels-up landing. This analysis addresses flightcrew performance, including the role of fatigue in the flightcrew’s performance, the adequacy of COA’s oversight of its pilots and the FAA’s oversight of COA, checklist design, and survival factors. 39 2.2 Flightcrew Performance Performance deficiencies exhibited by the flightcrew during this flight include: (1) failure to configure the hydraulic system for landing during the performance of the in-range checklist; (2) failure to detect initially that the flaps did not extend; (3) failure to determine the reason the flaps did not extend after detection; (4) failure to perform the landing checklist and to confirm the landing gear status; and (5) failure to discontinue the approach. 2.2.1 Failure to Properly Complete the In-range Checklist Fifteen minutes before landing, as the airplane descended through 19,000 feet, the captain omitted one item on the in-range checklist. The omitted item, “Hydraulics - ON & HI, CHECKED,” would have enabled the high pressure configuration of the hydraulic system, thereby providing pressure to operate the flaps and landing gear. Three steps were required to complete this checklist item: movement of the AUX and ALT pump switches from “OFF” to “ON,” movement of the left and right engine-driven hydraulic pump switches from “LOW” to “HI,” and confirmation that system pressures were between 2,800 and 3,100 psi.

[ANALYSIS] Consistent with other DC-9 pilots who reported failing to properly configure the hydraulic system, this crew detected a problem with flap deployment when the airplane did not respond with pitch and speed changes as flaps were selected to 15o and beyond. However, this crew did not recognize that the failure of the flaps to deploy was a symptom of improper hydraulic system configuration. Neither the captain nor the first officer recalled events concerning improper hydraulic system configuration in his previous DC-9 experience and, therefore, did not 41 At 0900:33, the captain said, “I think the flaps *.” At 0900:35, there were three intermittent sounds from the landing gear warning horn that, according to the first officer, were produced by the captain rapidly moving the throttles back and forth. The throttle manipulation would not have provided the captain with diagnostic information about the flaps. 44 possess firsthand knowledge to help recognize that the symptom he was experiencing was the result of this error. In addition, the Safety Board’s review of the information provided by COA to its pilots concerning the DC-9 hydraulic system revealed that the flight manual and training materials do not explicitly state that if the pumps are not switched to “HI,” the landing gear will not extend and the flaps will not deploy. The Safety Board concludes that the pilots’ lack of previous exposure, either through training or during line operations, to the consequences of improper hydraulic system configuration contributed to their failure to detect their hydraulic system configuration error. The Safety Board believes that the FAA should require all POIs of 14 CFR Part 121 operators using DC-9 and MD-80 airplanes with the “HI, LOW, OFF” hydraulic switch configuration to ensure that operating manuals and training programs include information about the consequences of improper hydraulic system configuration, specifically that the flaps and landing gear will not function normally if the engine-driven hydraulic pumps are not set to “HI.” 2.2.4 Failure to Perform Landing Checklist and Confirm Gear Position In accordance with company procedures, the first officer called for the landing checklist after the gear-down call. Although he placed the gear handle in the down position, the captain never initiated the checklist. Because the flaps remained stowed, the airplane did not slow during the approach. Traveling at a speed of approximately 200 knots, the airplane covered the distance between the outer marker and the runway threshold in about 75 seconds. If the target approach speed of 132 knots had been maintained, it would have taken about 115 seconds to cover this distance. The increase in speed allowed the flightcrew very little time to address the flap problem and configure the airplane for landing. In the 27 seconds that elapsed from the time the captain said “I think the flaps *” to the time the first officer stated “I don’t have flaps,” the captain manipulated the throttles and then responded to the gear-down call, the flaps 25 call, the flaps 40 call, and the flaps 50 call. The captain had very little time to react to the directives he was being given by the first officer.

[ANALYSIS] Therefore, it is unlikely that the norm contributed to the captain’s failure to properly configure the hydraulic system. However, the existence of informal norms, the findings from the line checks conducted after the accident, and the recent history of flightcrew-related accidents at COA raises concerns that an operational climate may have existed in which crews occasionally deviated from standard operating procedures. This significant issue is addressed in section 2.4 of this analysis. In summary, the Safety Board found no evidence indicating that the captain was interrupted or distracted during the performance of the in-range checklist, that the omitted checklist item was obscured, or that the captain believed the first officer would configure the hydraulic system. The Safety Board was unable to determine the specific reason for the captain’s omission of the “Hydraulics” item on the in-range checklist. The COA DC-9 Flight Manual states that both pilots are responsible for visual confirmation that all checklist items are completed. The first officer’s response to two items on the checklist, “Flight Instruments, Altimeters” and “Shoulder Harness,” indicates that he was aware that the in-range checklist was being completed. However, the first officer did not detect the captain’s omission of the “Hydraulics” item. The Safety Board concludes that because the captain omitted the “Hydraulics” item on the in-range checklist and the first officer failed to detect the error, hydraulic pressure was not available to lower the landing gear and deploy the flaps. The Safety Board is concerned that the normal in-flight operating procedure for the DC-9 hydraulic system deactivates (or, in the case of the MD-80, impairs the operation of) certain hydraulic components, including the landing gear and the flaps, without providing an overt signal to the flightcrew of the non-functional status of those components. If the hydraulic system is not configured properly during performance of the in-range checklist, the error can initially only be determined by direct observation of the hydraulic pump switches and pressure gauges. Because the flaps and landing gear are not typically extended until the later stages of an approach, the next opportunity for the flightcrew to detect such an error occurs during a period of higher workload when there is less time for problem diagnosis. The February 1995 ASRS report and statements by other DC-9 pilots indicate that failure to configure the hydraulic system for landing is not an uncommon occurrence.38 A review of checklists from several DC-9 and MD-80 operators revealed that none of the checklists, including the Douglas Aircraft Company’s checklist, emphasize the importance of the 38 See discussion in section 1.18.1. 41 “Hydraulics” item by placing it as the first item on the in-range checklist or requiring mandatory cross-check of the item by both pilots. Further, the item requires only an “ON & HI, CHECKED” response to the challenge and does not require verbal notation of the pressure values. NASA-sponsored research on checklist design39 indicates that critical items should be placed first on a checklist because the probability of successfully accomplishing the first items on a checklist is the highest.

[ANALYSIS] The only means available to fly the airplane was through manipulation of thrust available from the No. 1 and No. 3 engines. The primary task confronting the flightcrew was controlling the airplane on its flightpath during the long period (about 60 seconds) of the "phugoid" or pitch oscillation. This task was extremely difficult to accomplish because of the additional need to use the No. I and No. 3 power levers asymmetrically to maintain lateral (roll) control coupled with the need to use increases and decreases in thrust to maintain pitch control. The fl ightcrew found that despite their best efforts, the airplane would not maintain a stabilized flight condition. .·. I ,;; ·:: .. / ·,. 76 Douglas Aircraft Company, the FAA, and UAL considered the total loss of. hydraulic-powered flight controls_ so remote as to negate any requirement for an appropriate procedure to counter such a situation. The most comparable maneuver that the fl ightcrew was required to accomplish satisfactorily in a DC-10 simulator was the procedure for managing the failure of two of the three hydraulic systems; however, during this training, the remaining system.was available for movement of the flight controls. The CVR recorded the flightcrew's discussion of procedures, possible solutions, and courses of action in dealing with the loss of hydraulic system flight controls, as well as the methods of attempting an emergency landing. The captain's acceptance of the check airman to assist in the cockpit was positive and appropriate. The Safety Board views the interaction of the pilots, including the check airman, during the emergency as indicative of the value of cockpit resource management training, which has been in existence at UAL for a decade. The loss of the normal manner of flight control, combined with an airframe vi brat ion and the visual assessment of the damage by crewmembers, led the flightcrew to conclude that the structural integrity of the airplane was in jeopardy and that it was necessary to expedite an emergency landing. Interaction between the fl i ghtcrew and the UAL system aircraft maintenance network (SAM) did not lead to beneficial guidance. UAL flight operations attempted to ask the flightcrew to consider diverting to Lincoln, Nebraska. However, the information was sent through flight dispatch and did not reach the flightcrew in time to have altered their decision to land at the Sioux Gateway Airport. · The simulator reenactment of the events leading to the crash landing revealed that 1 ine fl ightcrews could not be taught to control the airplane and land safely without hydraulic power available to operate the flight controls. The results of the simulator experiments showed that a landing attempt under these conditions involves many variables that affect the extent of controllability during the approach and landing.

[ANALYSIS] The results of the simulator experiments showed that a landing attempt under these conditions involves many variables that affect the extent of controllability during the approach and landing. In general, the simulator reenactments indicated that landing parameters, such as speed, touchdown point, direction, attitude, or vertical velocity could be controlled separately, but it was virtually impossible to control all parameters simultaneously. After carefully observing the performance of a control group of DC-IO-qualified pilots in the simulator, it became apparent that training for an attempted landing, comparable to that experienced by UA 232, would not help the crew in successfully handling this problem. Therefore, the Safety Board concludes that the damaged DC-10 airplane, although flyable, could not have been successfully landed on a runway with the loss of all hydraulic flight controls. The Safety Board believes that under the circumstances the UAL flightcrew performance was highly commendable and greatly exceeded reasonable expectations. • . ' .-; ~ ~ .·-~· ! ! ' ,I; . ... :· 2.4 2.4.1 77 Analysis of Fan Disk Fracture Separation of Fan Disk Examination of the fracture surfaces of the fan disk disclosed that the near-radial, bore-to-rim fracture was the primary fracture. The fracture initiated from a fatigue region on the inside diameter of the bore. The remaining portions of the disk fractures were typical of overstress separations resulting from the fatigue failure. · Because of the geometry of the fan disk and the load paths within the disk, the near-radial fracture created a bending moment in the disk arm and web that overstressed the disk, leading to rupture and release of a segment. As soon as the segment of the disk was released, the remainder of the disk was immediately out of balance. Sufficient evidence in the form of witness marks,, on the containment ring indicates that the segment of the disk with its blade roots still attached exited the engine around the 7:30 position. Additional evidence from the bearing housings and compressor section indicates that the remainder of the disk with attached blade roots immediately exited the engine from about the 1:00 position. Blade fragments, separately and in groups, were primarily liberated toward the right horizontal stabilizer and the aft lower fuselage area. The investigation disclosed that the liberated pieces of the engine banjo frame contained transferred titanium. However, the Safety Board could not determine which of the titanium engine parts struck the frame. 2.4.2 Initiat'ion and Propagation of Fatigue Crack . Metallurgical examination showed that the fatigue crack initiated in a nitrogen-stabilized type I hard alpha defect at the inside surface of the bore.

[CONCLUSIONS > FINDINGS] I and No. 3 hydraul i c ~ system lines, and the forces of the engine failure fractured ~ the No. 2 hydraulic system, rendering the airplane's three hydraulic-powered flight control systems inoperative. Typical of all wide-body design transport airplanes, there are no alternative power sources for the flight control systems. 6. The airplane was marginally flyable using asymmetrical thrust from engines No. 1 and 3 after the loss of all conventional flight control systems; however, a safe landing was virtually impossible . 7. The airport emergency response was timely and initially effective; however, cornstalks on the airfield and the failure of the Kovatch P-18 water supply vehicle adversely affected firefighting operations. · 8. The FAA has not adequately addressed the issue of infant occupant protection. The FAA has permitted small children and infants to be held or restrained by use of seatbelts during turbulence, landing, and takeoff, posing a danger to themselves and others . l i ~ j I) ~I :o ! 1 l ; I 1.- 1 l l l I j l I .i j ~1 0 }} l I J I l 1 l j 1 l i I ~'I 9 1 l . l 101 0 9. · ··Separation of the titanium alloy stage 1 fan rotor disk was · ·.··.the result of a fatigue crack that initiated from a type 1 hard alpha metallurgical defect on the. surface ·of the disk bore. 10. The ·hard alpha metallurgical defect was formed in .·.the titanium alloy material during manufacture of the ingot from which the disk was forged. 11. The hard alpha metallurgical defect was not detected by ultrasonic and macroetch inspections performed by General Electric Aircraft Engines during the manufacturing process of the disk. 12. The metallurgical flaw that formed during initial manufacture ··of the titanium alloy would have been apparent if the part had ·been macroetch inspected in its final part shape. 13. The cavity associated with the hard alpha metallurgical defect was created during the final machining and/or shot peening at the time of GEAE's manufacture of the disk, after GEAE's ultrasonic and macroetch manufacturing inspections. 14. The hard alpha defect area cracked with the application of stress during the disk's 1nitia1 exposures to full thrust engine power conditions and the crack grew until it entered material unaffected by the hard alpha defect. 15. General Electric Aircraft Engines material and production records relevant to CF6-6 stage 1 fan disk S/N MPO 00385, which was the failed disk, were incomplete. 16.

[ANALYSIS] The location of parts of the No. 2 engine and empennage structure near Alta, Iowa, together with the ' 75 documentation and analysis of the No. 2 engine components and surrounding structure, led the Safety Board to conclude that the No. 2 engine stage 1 fan disk fracture and separation was the initial event that led to the liberation of engine rotating parts with sufficient energy to penetrate the airplane's structure. · · Shortly after the engine failure, the crew noted that the hydraulic fluid pressure and quantity had fallen to zero in the three systems. Approximately l minute after the engine failure, the FDR recorded no further powered movement of the flight control surfaces. Consequently, the No. 2 engine failure precipitated severe damage that breached the three hydraulic systems, leaving the flight control systems inoperative. Titanium alloy was found on the fracture surfaces of severed lines of hydraulic systems No. 1 and No. 3 located in the right horizontal stabilizer. Several of the major components of the engine, including the stage I fan blades and fan disk, were made from titanium alloy and no other components of the surrounding airframe were made from such material. These factors led the Safety Board to conclude that the systems' No. 1 and No. 3 hydraulic lines were severed by fragments released during the failure sequence of the No. 2 engine. The loss of hydraulic system No. 2 required further analysis. The engine-driven No. 2 hydraulic pumps were attached to and received power from the No. 2 engine accessory section. This unit was mounted to the engine directly below the fan section of the engine. Portions of the No. 2 engine accessory section and associated No. 2 hydraulic system components, including hydraulic supply hoses, were found in the Alta, Iowa, area. Therefore, portions of the No. 2 hydraulic system and supply hoses mounted on, or adjacent to, the No. 2 engine accessory section were damaged and separated by the forces and disruption of the engine fan section during the engine failure. The investigation disclosed no evidence of other system anomalies that would have contributed to the hydraulic system or flight control difficulties experienced in the accident. 2.3 Performance of UAL 232 Flightcrew Because of the loss of the three hydraulic systems, the flightcrew was confronted with a unique situation that 1 eft them with very limited control of the airplane. The only means available to fly the airplane was through manipulation of thrust available from the No. 1 and No. 3 engines. The primary task confronting the flightcrew was controlling the airplane on its flightpath during the long period (about 60 seconds) of the "phugoid" or pitch oscillation.

[ANALYSIS] During the _UA 232 accident sequence, once the fan disk failed and the pieces began to escape the confines of the containment ring, the dispersion of rotor disk and fan blade fragments was altered by contact with 90 both engine components and the airplane structure. The Safety Board did not attempt to determine the specific origin or trajectory of each fragment that damaged the airplane in flight. For accident prevention purposes and in the course of making safety recommendations, it was sufficient to recognize that catastrophic damage from the failure of rotating parts can originate from any fragment source with sufficient energy to penetrate the airplane's structure. The s·afety Board considers in retrospect that the potential for hydraulic system damage as a result of the effect of random engine debris should have been given more consideration in the original design and certification requirements of the OC-10 and that Douglas should have better protected the critical hydraulic system(s) from such potential effects. As a result of lessons learned from this accident, the hydraulic system enhancement mandated by AD-90-13-07 should serve to preclude loss of flight control as a result of a No. 2 engine failure. Nonetheless, the Safety Board is concerned that other aircraft may have been given similar insufficient consideration in the design for redundancy of the motive power source for flight control systems or for protecting the electronic flight and engine controls of new generation aircraft. Therefore, the Safety Board recommends that the FAA conduct system safety reviews of currently certificated aircraft in light of the lessons learned in this accident to give an· possible consideration to the redundancy and protection of power sources for flight and engine controls. 2.7.2 Future Certification Concepts On March 9, 1988, the FAA issued AC 20-128, in part as the result of a Safety Board recommendation made in 1982. The AC provides for a method of compliance with FARs that require design precautions to be taken to minimize the hazards to an airplane in the event of an uncontained engine or auxiliary power unit failure. The AC defines dispersion angles for fragments that may be released during a fan blade or rotor failure. These angles define impact areas relative to the engine installation based on recorded observations of the results of failures both in service and in tests. The AC also provides a listing of design considerations to minimize damage to critical structural elements and systems in the airplane, and defines the fragment energy levels that can be expected from the failure of a fan blade or predicted pieces of a rotor. The Safety Board notes that the AC provides the engine/airframe designer with information that had previously been left to the interpretation of the designer. The Safety Board also notes that the initial operational capability of the high-bypass-ratio turbofan engines began in the early 1970's. For almost 20 years, and obviously during the development period of the majority of the wide-body fleet, a recognized . interpretation· of the regulations concerning hazards related to uncontained engine failures was not published by the FAA.

[ANALYSIS] N1O0ME was certified, maintained, and equipped in accordance with applicable FAA regulations and approved procedures. The original airplane certification process had required demonstration of relevant handling qualities of the airplane, including conditions normally encountered in the event of sudden loss of thrust of either engine. The results of this investigation did not reveal any handling characteristics of the DC-9-14 which were inconsistent with the original standards for certification of the airplane. For example, the pilots who participated in the Safety Board's DC-9-14 flight demonstration deseribed the airplane's handling characteristics as docile, even after the sudden and complete loss of thrust from the right engine in a simulated takeoff/climb phase of flight. Consequently, the Safety Board concludes that the loss of control of the airplane was not directly attributable to the loss of thrust from the right engine. The analysis of this accident thus examined those factors which, in conjunction with, the failure of the airplane’s right engine, might have caused the pilots to lose control. Those factors included: ° The possibility that fragments of the right engine separated with sufficient energy and trajectory to cause critical damage to the airpiane’s flight control system; ~47T- ° The possibility of control system malfenction(s) which, in combination with a single or dual power loss, could have rendered the airplane uncontrollable; ° The possibility of a mechanical failure of the left engine, either related or unrelated to the failure of the right engine, which left the airpiane with insufficient thrust to maintain flight; and fe) The possibility of inappropriate flighterew response to the emergeney presented by the failure of the right engine. To resolve the factors which precipitated the loss of control, it was first necessary to examine the circumstances of the failure of the right engine. 2.2 ight ine Failure and Demage from Uneontained ine Parts The physical damage to the engine and the condition of the inlet fan biades and low pressure compressor blades indicated that the right engine had little or no rotation at impact with the ground. The sound spectrum examination of the CVRrecorded engine sounds also indicated that this engine lost rom very rapidiy after the engine failure. The hole in the high pressure compressor in the plane of rotation of the 9-10 Stage removable sleeve spacer and the damage to the compressor and spacer revealed conclusively that the spacer had ruptured in flight and that the spacer parts were not eontained by “he engine easing. The ejected spacer parts had ruptured the rear skirt intermediate .ase at the 11 to 1 o'clock position, leaving a 4-by 7-inch opening in the top of the case. The loss of the spacer and consequential Camage within the right engine caused a rapid deceleration and a complete loss of thrust from that engine.

[ANALYSIS] Also, it was found that the rudder hydraulic actuator, which controls rudder movement by hydraulic pressure or by transferring control input to the aerodynamic tab, showed no evidence of preimpact damage. Additionally, the rudder power shutoff vaive was found with a bent control rod and discoloration, consistent with rudder hydraulic power on at ground impact. The Safety Board also examined the possibility that the right engine cowl was biown open in flight or became distorted to such an extent that excessive drag was produced, affecting controllability of the aircraft. Aithough the right engine upper eowling was extensively damaged by impact forces, all four outboard iatches remained latched. There was no evidence to indicate that the right cowl had opened in flight. A recovered right cowl pieces which could be positively identified were found within the impact area. Although a small (2- by 2-inch) piece of metal whieh resembled cowl material was found near runway 19R, it was determined that each square foot of deformed cowling would produce drag equivalent to a reduction of engine thrust by 160 pounds--a minor factor. Based upon the small hole (4- by 77-inch) found in the right engine case, the absence of other case deformation (other than impact damages), and the characteristics of typical uncontained engine pieces ejected at high velocity, the Safety Board concludes that the cowling deformation probably was small and therefore caused very little additional drag following the right engine failure. 2.3 Flight Control System Failure or Malfunction The Safety Board considered the possibility of a flight control system failure or malfunction, unrelated to the right engine failure, that might have occurred simultaneously or nearly simultaneously with the right engine failure, and thet subsequently led to the loss of control. The Safety Board does not believe that such a failure or malfunction ceeurred for several reasons, ineluding those reasons cited previously regarding possible damage caused by the ri¢ht engine failure. In addition, an analysis of the contro! movements, which woukl have been required (commanded or etherwise) for the airplane to have maneuvered as indicated by the FDR, revealed that: 100% 80% DEBRIS 55% ENERGY (PERCENT) 0 ies ean en ee fi a I oococecocaccacdgeucesosazags ~*~ 06S, Figure 7.--Three-view drawing of a DC-3 series 10 airplane with various debris patterns and corresponding percentage of energy ioss. ~5i- (1) Rudder deflection to the left was required for the airplane to maintain heading for 4 seconds immediately after the right engine fejlure; (2) Rudder deflection to the right was required to cause the heading ehenge which occurred from the 4th to the 10th second after the right engine failure; (3) Elevator control was required to cause the pitcn-over and pull-up maneuvers which were documented by the FDR acceleration traces after the right engine failure; and (4) Aileron/spoiier deflection was required for the airplane to maintain the roll attitude in the presence of large sideslip angles which were docu mented.

[ANALYSIS] The crack had propagated to a length which should have sllowed detection on the occasion of the last high pressure compressor overhaul and spacer rework in 1981. 8. None of the airplane flight control systems were disabled. 9. The cause cf the left engine power loss, which occurred beginning about 1.5 seconds after the right engine failed, was not determined. 19. The left engine experienced a compressor stall in the last seconds of the flight after contre] had been lost and the airplane was descending toward the ground in an unusual attitude. 11. The loss of teft engine power was not significant with respect to the loss of control of the airplane. 12. 13. 14, 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. ~68- The captain initially responded correctly with deflection of the rudder pedal to the left to compensate for the loss of right engine thrust and by lowering the nose of the aircraft; however, he appeared to be unaware of the exact nature of the emergency. The crew response to the right engine failure was not coordinated. Neither pilot verbally identified the emergency condition or made the emergency callouts required by FAA-approved Midwest Express procedures. The rudder was incorrectly deflected to the right 4 to 5 seconds after the right engine failure. An accelerated stall and loss of control occurred 10 seconds after the failure of the right engine. Forward visual cues (outside the cockpit) were nut available to the crew at the time that the right engine failed. Peripheral visual cues were available. The visual flight simulator, which wes used by the crewmembers in training, did not provide onset y-w aad longitudinal acceleration cues, peripheral visual cues, or aural cues which were available to the crew in the airplane. The captain and first officer misinterpreted the inside visual cues which were presented in the airplane. The differences in visual motion and aural cues presented in the visual flight simulator and in the airplane may have limited the ability of the flighterew to recognize and react ar-oropriately to the emergency. Failure to recognize the nature of the emergency and improper operation of flight controls precipitated the loss of control. The DC-9-14 does not require unusual pilot skill or strength to maintain continued flight foliowing an engine failure on takeoff. Both crewmembers were relatively inexperienced in DC-93 flight operations. The FAA Principal Operations Inspector who was responsible for oversight of Midwest Express was inexperienced in FAR 121 turbojet air carrier operations. A "silent cockpit" philosophy was suggested by Midwest Express in response to certain emergency situations, although the concept was not approved by the FAA and was in conflict with approved emergency procedures.

[ANALYSIS] The Safety Bogrd also considered the possibility that the captain's left rudder pedal support arm fractured after he deflected the left rudder pedal. Even though the fracture mode of the pilot's pedals cowld not be determined conclusively from metallurgical examination, there was no evidence that a pedal failed during a critical phase of flight. Review of CYR sounds revealed no noise or flighterew response that could be associated with such en event. The similarity of the fractures on both the captain's and copilot's pedals suggests that they were subjected to similar forces at failure, most probably overstress at impact. Furthermore, failures of rudder pedals in past incidents oceurred during braking actions while on the ground, rather than during inflight operation because rudder control forces applied in flight produce less stress on the pedals. Finally, failure of the left rudder pedal would result in return of the rudder to a near-neutral position, and would not account for the deflection of the rudder to the right. Therefore, the Safety Board concludes that the rudder pedal support arm fractures were eaused by overstress forces at impact and were not related to the cause of the accident. In conclusion, the Safety Board determined, after examination of the wreckage, trajectory study calculations, and research into the DC-9-14 control system, that engine parts probably did not strike the aircraft after being ejected fram the right engine. The Board believes that if any of the smali engine parts actuallv struck the aircraft, no damage of any consequence would have oecurred as a result of that contact. Tne Board found that there was no basis upon which to conelude that flight control systems malfunctioned or were damaged in flight, secondary ts the right engine failure, and that ali of the onboard flight control systems on NIOQOME were available to the fligntcrew following the abrupt loss of right engine power. Further, electrical power was available to the crew of flight 105 until impact based upon the continuous operation of the FDR and the CYR and the elongation of the right wingtip navigation light buib filament, which indicated that the filament was hot and, therefore, on when subjected to impact forces. 2.4 Left ine Power Loss The Safety Board does not believe that the left engine power loss was signifieant with respect to the eventual loss of control of the airplane. Any reduction in left engine thrust that occurred before stickshaker would have reduced the yawing of the airplane which occurred after the right engine failure. While possibly necessitating a foreed landing. a left engine power ioss should not have precipitated, or even contributed to, a loss of control of the airplane. However, the reduction in ieft engine power could have confused the crew.

[ANALYSIS] Returning the rudder to neutral and holding neutral rudder, after initially applying rudder to correct for differential engine thrust. wouid not have created the heading change rates which were indicated by the FDR data. Likewise, a system malfunction which would cause the rudder to trail in a near-neutral position would be inconsistent with the FDR-indicated heading change data. The demonstration flight in a DC-9-14 airplane showed that the airplane had no control characteristics which were inconsistent with the applicable certification standards; the airplane was feund to be fuy controliable in an engine-out flight environment, even without using rudder (the primary control for correeting vaw and maintaining heading) to correct for yaw. Having found no evidence or airplane performance basis for concluding that there was a control system failure or malfunction, the Safety Board concludes that the rudder deflection, which occurred beginning 4 to 5 seconds after the right engine failure, was the result of the flightcrew's improper response. Based on the analysis of the airplane performance, the yaw generated by the incorrect rudder deflection, combined with G loading, caused the airplane to enter an accelerated stall at an altitude tco low for recovery. In the seconds which preceded the accelerated stall and loss of eontrol, the airpiane was ina very dynamic situation. The increasing rate of roll, the sideslip, and the increase in acceleration load ali affected adversely the stall speed. Because of the rapidly changing attitude of the airplane, the pilots would not have been expected to know the speed at which the airplane would stall in aecelerated flight. Compared to the increase in stall speed, the 8-Knot error in indicated airspeed (due to static source error in a sideslip) would not have been significant. Further, the stickshaker stall warning system would not and did not provide the customary 4 to 5 seconds warning which is typical of that system because of the rapid entry into the stall. The Safety Board concludes that the stall occurred because the flighterew dic not diagnose the nature of the emergency correctly, applied ineorreet rudder control about 4 to 5 seconds after the right engine failure, and applied nose-up elevator control which increased the G loads. The nose-up elevator contro}} input would have been a normal response to correct for the pitech-over maneuver and the reduction in pitch attitude which was precipitated Dy the rudder pedal induced roll and was consistent with the rapid deceleration of the airplane. The rapid deceleration would have resulted in a vestibular perception of downward pitching of the nose of the airplane. The Safety Board believes that more effective scanning of the flight and engine instruments by the pilots of fight 105 would have enabled them to maintain control of the airolane and to properly evaluate the powerplant anomalies.

[ANALYSIS] Both crewmembers were relatively inexperienced in DC-93 flight operations. The FAA Principal Operations Inspector who was responsible for oversight of Midwest Express was inexperienced in FAR 121 turbojet air carrier operations. A "silent cockpit" philosophy was suggested by Midwest Express in response to certain emergency situations, although the concept was not approved by the FAA and was in conflict with approved emergency procedures. FAA surveillance of Air Carrier Engine Service (AeroThrust) was ficient in the 2-year period which preceded the overhaul of the 9-10 spacer. 27. The accident was nonsurvivable because the impact forces exceeded the limitations of human tolerance. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of the accident was the flignterew's improper use of flight controls in response to the catastrophic failure of the right engine during a critical phase of flight, which led to an accelerated stall and loss of control of the airplane. Contributing to the loss cf control was a lack of crew coordination in response to the emergency. The right engine failed from the rupture of the 9th to 16th stage removable sleeve spacer in the high pressure compressor because of the spacer’s vulnerability to cracks. 4. RECOMMENDATIONS On November 8, 1985, the Safety Board reeommended that the Federal Aviation Administration: A~895-120 Issue an Airworthiness Directive (AD) to require the installation of the one-piece, integral sleeve spacer at ali six locations in the high-pressure compressor rotor of Pratt & Whitney dJT8D-series engines not so equipped. The installation should be made as soon as practical but no later than the next opportunity wherein the engine is available in tr - maintenance facility where a partial or complete disassembly of the eompressor can De accomplished. A~85-121 Notify sppropriate foreign civil aviation authorities and foreign operators of airplanes equipped with Pratt & Whitney JT8D-series engines of the failures associated with the removable sleeve spacers instalied in the high-pressure compressor rotor and of the actions which should be taken to minimize or eliminate the failures. On April 7, 1986, the Safety Board recommended that the Federal Aviation Administration: A~86 ~28 Issue a Telegraphic Airworthiness Directive and amend the airworthiness directive proposed in the Notice of Proposed Rulemaking published at 51 FR 37, Docket No. 85-ANA-46, to require that the one-time, on-wing eddy current inspection specified in the proposed airworthiness directive be repeated at 1,000-cycle intervals until stage 7-8, 8-9, and 9-i0 removable sleeve spacers between the high-pressure compressor are replaced with integral sleeve spacers. As a result of its investigation, the Safety Board recommended that the Federal Aviation Administration: Issue an air carrier operations bulletin directing Principal Operations Inspectors to review their respective air carrier's flighterew training programs to ensure the existence of new coordination procedures that, -70.

[ANALYSIS] The numerous references in the draft report to the recoverability of the airplane reveal another “backdoor” through which the NTSB arrives at its probable cause determination. Presumably, the NTSB’s logic is that if an airplane suffering a dual PCA jam can be recovered in the simulator, then the failure to recover Flight 990 means that it did not experience a dual PCA jam. The obvious flaw in this logic is that no matter how good the simulator is, it cannot reproduce the feelings, sensations, and emotions triggered by an unexpected upset, at night, over the ocean, during which the airplane and its contents experience a 40 degree nose down pitch and the crew and passengers experience negative g forces. Further, in none of its references to “recoverability” does the NTSB ever define what it means by that term. If recovery were defined as a point at which an uncommanded descent has been arrested, the wings are level, and the pitch has almost reached zero, then the Flight 990 33 crew recovered the airplane because those were the flight characteristics when the FDR stopped recording. Precisely what happened after Flight 990 recovered is unknown because both the CVR and FDR stopped recording. It does appear, however, from the location and condition of the left engine, that pieces of the airplane, including the left engine, departed the airplane prior to the beginning of the second, fatal dive toward the ocean. The loss of the left engine -- likely due to an overstress of the airplane during the ascent back to 24,000 feet -- created an asymmetrical load on the airplane from which recovery was virtually impossible. Consequently, regardless of the reason that the airplane initially departed cruise flight, there is a substantial argument that the reason that Flight 990 crashed was because it lost a major component -- the left engine -- which rendered the airplane unstable and unrecoverable. 26. (Page 58) As a result of the EgyptAir’s January 12, 2001 response to the Boeing submission, the NTSB was compelled to conduct further simulator tests and bench tests of the pressure relief valves of the elevator control assembly. These additional tests were required because of the incorrect data -- identified by EgyptAir -- that Boeing had submitted to the NTSB. The inaccuracies in the Boeing submission resulted in incorrect blowdown data for the elevator which was subsequently used as the basis for some of the airplane performance analysis. These facts should be included in the report to document the inaccuracies of the Boeing performance data relied upon by the NTSB. The variances in Boeing data demonstrate how easy it is to move the FDR flight profile towards, or away from, the E Cab simulator profile for a dual actuator failure. This aspect of the investigation underscores the manner in which the factual record and the subsequent analysis was selectively developed in the draft report. 27. (Page 59) The NTSB’s assertion that there was “no evidence of any ATC problems or issues” is incorrect.

[ANALYSIS] However, those simulations assumed that there were no opposing pilot inputs. The captain’s failure to recover the airplane can be explained, in part, by the relief first officer’s opposing flight control inputs. It is possible that efforts to recover the airplane after the airplane lost electrical power were also complicated by the loss of electronic cockpit displays. In summary, the evidence establishes that the nose-down elevator movements were not the result of a failure in the elevator control system or any other airplane system but were the result of the relief first officer’s manipulation of the airplane controls. The evidence further indicates that the subsequent climb and elevator split were not the result of a mechanical failure but were the result of pilot inputs, including opposing pilot inputs where the relief first officer was commanding nose-down and the captain was commanding nose-up movement. The Safety Board considered possible reasons for the relief first officer’s actions; however, the Board did not reach a conclusion regarding the intent of or motivation for his actions. 118 This sentence, “Get away in the engines,” is an example of a phrase where direct translation of the Arabic words into English with no attempt to interpret or analyze the words resulted in an awkward or seemingly inappropriate phrase. In this case, it is possible that the captain, surprised to realize that the engines had been shut off, was trying to tell the relief first officer to leave the engines alone. However, research indicates that poor word choice, improper grammar, and the use of incomplete phrases can be symptomatic of high levels of psychological stress in a speaker. 66 NTSB/AAB-02/01 Summary 1. The accident airplane’s nose-down movements did not result from a failure in the elevator control system or any other airplane failure. There was no evidence of any failure condition within the elevator system of the accident airplane that would have caused or contributed to the initial pitchover or prevented a successful recovery. No mechanical failure scenario resulted in airplane movements that matched the flight data recorder data from the accident airplane. Even assuming that one of the four examined failure scenarios that the investigation evaluated in depth had occurred, the accident airplane would still have been recoverable because of the capabilities of the Boeing 767’s redundant elevator system. 2. The accident airplane’s movements during the initial part of the accident sequence were the result of the relief first officer’s manipulation of the controls. At the relief first officer’s suggestion, a transfer of control at the first officer’s position occurred earlier than normal during the accident flight. The relief first officer was alone in the cockpit when he manually disconnected the autopilot and moved the throttle levers from cruise to idle; there was no evidence of any airplane system malfunction, conflicting air traffic, or other event that would have prompted these actions. The nature and degree of the subsequent nose-down elevator movements were not consistent with those that might have resulted from a mechanical failure but could be explained by pilot input. There was no apparent reason for the relief first officer’s nose-down elevator inputs.

[ANALYSIS] However, the draft report does not address these issues and simply dismisses the possibility that any of the failure scenarios are consistent with the accident aircraft FDR data. Again, the NTSB’s failure to include any discussion of the forces necessary to move the flight control surfaces creates a 37 misleading impression. The NTSB’s claim that “the nonfailed surface would have responded immediately to any nose-up flight control inputs” is not correct. The nonfailed surface would not respond to “any” nose-up input; it would only respond to nose-up input of a sufficient force to overcome the failed condition. As demonstrated in the ground testing, this force is considerably higher than the normal force. Whether the crew would immediately apply sufficient force to overcome a failure for which they were not trained is a matter of conjecture. This same comment applies to the NTSB’s wholly conclusory assertion that the lack of demonstrated nose-up movement proves that there was no attempt to recover the airplane (see p. 66). There is no evidence whatsoever to support this statement, which is made solely for the purpose of placing blame for the accident on the RFO. The fact is that the RFO could have been attempting a recovery, but failed -- or was not able -- to exert sufficient force to move the elevators in a failed condition. The question of the RFO’s ability to exert the necessary force is significant when Figure 4 is examined. Figure 4 shows that the downward deflection of the elevators tracked exactly the decrease in load factor. This may indicate that as the load factor decreased, the RFO became less able to exert the force necessary for recovery and, thus, the force of the PCA failure overcame the force of recovery. Figure 4 also shows that when the load factor began to increase, the elevators also began to move TEU -- possibly in response to the increasing ability of the crew to gain the leverage necessary to exert recovery level forces. 35. (Page 67) The draft report seriously understates the limitations of the E Cab simulator and the Boeing aircraft performance data with the simple statement that “the Board is well prepared to take those limitations into account … In this case the Board determined that the differences were not significant ….” This sweeping conclusion again indicates a preconceived conclusion by the Safety Board to not involve the airplane in the accident scenario. More importantly, the draft report omits any discussion explaining exactly how the NTSB took the simulator limitations into account, what differences were considered, and why those differences were deemed “not significant” in the context of this investigation. 38 36. (Pages 67-68) As part of its conclusion that there was no mechanical failure, the NTSB argues that the RFO “exhibited no audible expression of anxiety or surprise.” This fact -- even if true -- does not mean that there was no PCA jam or that the elevator bellcrank rivets did not shear. By making this argument in the mechanical defect analysis section of the draft report, the NTSB illustrates the biased and predetermined nature of the report.

[ANALYSIS] The Safety Board also conducted simulations in which pilots from Boeing, EgyptAir, the Federal Aviation Administration (FAA), and the Board evaluated the controllability of the airplane following an initial upset that might have been caused by any of these failure scenarios. During these simulations, the pilots were consistently able to regain control of the airplane and return it to straight and level flight using normal piloting techniques, and the airplane could be trimmed to hands-off level flight. In fact, the 767’s redundant actuation system is designed to allow pilots to overcome dual failures such as these. Even though increased control forces were necessary, recovery could be accomplished by a single pilot using either the left or right control column.102 Further, the simulations also demonstrated that the airplane could climb to about 25,000 feet msl with the engines shut down, even with the speedbrakes extended. The simulation also documented that the engines could have been promptly restarted and (assuming there were no opposing pilot inputs) that the airplane could have been recovered during the climb after the recorders stopped recording. Although the Safety Board recognizes that the simulator did not duplicate the accident airplane’s actual flight conditions in every way,103 such limitations are not uncommon in simulations, and the Board takes those limitations into account when evaluating simulator results. In this case, the Board determined that the differences were not significant and did not affect the validity of the results of the simulations. Immediately after the airplane’s initial nose-down dive, the relief first officer would have felt an immediate uncomfortable sensation as the airplane’s load factor decreased to near 0 Gs. He should also have noted sudden changes in the airplane’s pitch attitude, pitch rate, airspeed, and altitude. In response to these obvious cues, the relief first officer did not attempt to counter the dive by commanding nose-up elevator, a largely intuitive pilot response to initiate a recovery. 101 Further, although the first three failure scenarios evaluated in depth involved simultaneous dual PCA failures at the start of the accident sequence, as previously discussed, it is also clear from the FDR data that no latent jam of a single PCA occurred before the accident sequence. 102 As a former chief flight instructor with 5,191 hours in the 767, the relief first officer should have been readily able to regain control of the airplane. 103 For more information about the limitations of Boeing’s simulator, see the section titled, “Potential Causes for Elevator Movements During the Accident Sequence.” 58 NTSB/AAB-02/01 Nor did the relief first officer exhibit any audible expression of anxiety or surprise or call for help during the airplane’s initial dive or at any time during the remainder of the recorded portions of the accident sequence. Further, the relief first officer did not respond to the captain’s repeated question, “What’s happening?” after the captain returned to the cockpit.

[ANALYSIS] Mechanical Failure/Anomaly Scenarios The Safety Board evaluated possible mechanical failure and pilot action scenarios in an attempt to determine whether they were consistent with the elevator movements made during the accident sequence. As previously discussed in the section titled, 87 The engineering simulator was modified to model the left and right elevator surfaces independently, and, using split elevator movements similar to those recorded on the FDR, the simulator was able to duplicate the FDR-recorded pitch history. 54 NTSB/AAB-02/01 “Potential Causes for Elevator Movements During the Accident Sequence,” the investigation ruled out all but four possible anomalies and failure scenarios as potential factors in the accident because they diverged too far from what was reflected on the accident flight’s FDR to warrant further consideration.88 Analysis showed that the effects of four failure scenarios (each of which involves dual failures) bore some resemblance to some portions of the accident flight’s FDR data. Specifically, initially it appeared that each of these failure scenarios could potentially cause nose-down elevator movements or a split elevator condition that might resemble those recorded on the accident flight’s FDR. Those four failure scenarios were (1) disconnection of the input linkages to two of the three PCAs on the right elevator surface,89 (2) a jam of the input linkages or servo valves in two of the three PCAs on the right elevator surface,90 (3) a jam of the input linkage or servo valve in one PCA and the disconnection of the input linkage to another PCA on the right elevator surface,91 and (4) a jam in the elevator flight control cable connecting the right-side control column to the right aft quadrant assembly combined with a break in the same cable.92 Therefore, the wreckage from the accident airplane was examined for possible evidence of PCA anomalies, and the predicted elevator movements resulting from these failure scenarios were evaluated and compared with the data from the accident flight. As previously mentioned, one of the recovered PCAs was found with a pin (that attached the spring guide to the servo valve slide) sheared and one coil of the bias spring improperly positioned over the head portion of the spring guide.

[ANALYSIS] Given the documented safety problems with the elevator control system and the ongoing, unresolved deficiencies, the dismissive treatment by the NTSB 27 in the draft report is surprising. This lack of emphasis potentially could be justified if the cause of the sheared rivet failures were known and had been thoroughly analyzed. However, without knowing the cause of the failures, the NTSB cannot dismiss the issue as unrelated to the EgyptAir Flight 990 accident and still expect to maintain high standards of accident investigation credibility. 12. (Page 25) This section states that the airplane is controllable with a dual bellcrank failure. However, there is no discussion of the increased control forces a pilot would experience or the fact that there is no pilot training for this failure. This information is critical to a proper analysis of pilot actions. 13. (Page 30) Speech Pattern Information -- This entire section was developed without consideration to how a person would react if he were so shocked or ill-prepared by an unexpected airplane system failure that he reverted to a state of inaction. This behavior has been documented by the NTSB in the past, but is not accounted for here. 14. (Page 32) The purportedly “unintelligible” comment was interpreted by some in the CVR group as “control it.” If “control it” was uttered by a crewmember in response to observed, unexplained, movement of the control column or the control wheel, it explains both the comment and the decision to disconnect the autopilot. Again, this comment is not included or discussed in the draft, especially within the context of the statement of the EgyptAir captain who observed this activity on the outbound flight. It is puzzling that the NTSB repeatedly describes this phase as “unintelligible” when five of the nine CVR Group members were able to understand the words -- with four believing it to be “control it” and one thinking it was “hydraulic.” In fact, it was only a minority of the group that found the speech “unintelligible.” In light of these facts, it is wrong for the NTSB to label the utterance as unintelligible. The comment that these words “might have been an isolated declaration” of the RFO is sheer speculation. In fact, given the RFO’s reported lack of comfort using English, it is highly unlikely that he was the speaker. 28 15. (Page 34) Here, the NTSB describes its review of the transcripts of other accidents and states that it “did not observe instances on these CVR recordings in which a pilot failed to exhibit surprise, alarm, or increased stress . . . .” The NTSB’s purpose in making this comparison with Flight 990 obviously is to suggest that the absence of exclamations from the RFO indicates he was in control of the events and, therefore, was not surprised.

[CONCLUSIONS > FINDINGS] NTSB Aircraft Accident Report 54 11. The flight nurse in the left aft seat had likely been restrained in his seat and was likely ejected from the helicopter with his seat during the accident sequence. 12. The impact forces of this accident were survivable for the helicopter occupants. 13. If the helicopter had been equipped with a crash-resistant fuel system, the potential for thermal injuries to the occupants would have been reduced or eliminated. 14. Those who purchase, lease, or contract for helicopter services and those who operate or fly aboard helicopters as part of their job are likely unaware that the designs of most existing and newly manufactured helicopters do not include the improved crashworthiness standards required of newly certificated helicopters, which could compromise occupant protection if an accident were to occur. 15. Data to better understand the safety issues involved in this accident could likely have been recovered from a flight recorder system that complied with the provisions of Federal Aviation Administration Technical Standard Order C197, “Information Collection and Monitoring Systems.” 2.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was Airbus Helicopters’ dual-hydraulic AS350 B3e helicopter’s (1) preflight hydraulic check, which depleted hydraulic pressure in the tail rotor hydraulic circuit, and (2) lack of salient alerting to the pilot that hydraulic pressure was not restored before takeoff. Such alerting might have cued the pilot to his failure to reset the yaw servo hydraulic switch to its correct position during the preflight hydraulic check, which resulted in a lack of hydraulic boost to the pedal controls, high pedal forces, and a subsequent loss of control after takeoff. Contributing to the accident was the pilot’s failure to perform a hover check after liftoff, which would have alerted him to the pedal control anomaly at an altitude that could have allowed him to safely land the helicopter. Contributing to the severity of the injuries was the helicopter’s fuel system, which was not crash resistant and facilitated a fuel-fed postcrash fire. NTSB Aircraft Accident Report 55

[CONCLUSIONS > FINDINGS] As a result of the findings from that investigation, the NTSB reiterated Safety Recommendation A-13-13 on October 23, 2014. NTSB Aircraft Accident Report 53 2. Conclusions 2.1 Findings 1. The helicopter was properly certificated, equipped, and maintained in accordance with federal regulations. None of the available evidence indicated any preimpact structural, engine, or system failures. 2. The pilot was properly certificated and qualified in accordance with federal regulations. Pilot fatigue and the pilot’s medical conditions and prescribed medications were not factors in this accident. 3. The wind conditions at the time of the accident would not have prevented the pilot from maintaining yaw control of the helicopter. 4. The pilot most likely did not return the yaw servo hydraulic switch to its correct (“ON”) position before takeoff, resulting in a lack of hydraulic pressure to the tail rotor servo control and the yaw load compensator accumulator. 5. A lack of hydraulic boost to the pedals, resulting in significantly increased pedal loads, was the most likely cause of the loss of tail rotor control, which led to the left yaw that occurred simultaneously with takeoff. 6. A salient alert for insufficient hydraulic pressure in the tail rotor hydraulic circuit could have cued the pilot to the incorrect configuration of the tail rotor hydraulic circuit, the lack of hydraulic boost to the pedal controls, and the resulting increased pedal loads. 7. Although not required to do so, Air Methods did not aggressively take action to comply with Airbus Helicopters’ Service Bulletin No. AS350-67.00.64, which called for installing a flashing light on the cockpit caution and warning panel to alert pilots that the yaw servo hydraulic switch was in the incorrect position; if this nonmandatory service bulletin had been complied with, the pilot might have noticed that the switch was not in the correct (“ON”) position before takeoff. 8. The design of Airbus Helicopters dual-hydraulic AS350-series helicopters did not account for the possibility of pilot error in configuring the tail rotor hydraulic circuit or assessing the functionality of the yaw load compensator, and efforts to address these safety issues have thus far been insufficient. 9. Despite the significantly increased pedal loads, the pilot continued the takeoff to climb the helicopter above nearby obstacles and gain forward airspeed to counter the left yaw rotation, but his efforts were unsuccessful. 10. If the pilot had performed a hover check, he would have identified the pedal control anomaly at an altitude that could have afforded a safe landing on the helipad. NTSB Aircraft Accident Report 54 11. The flight nurse in the left aft seat had likely been restrained in his seat and was likely ejected from the helicopter with his seat during the accident sequence. 12. The impact forces of this accident were survivable for the helicopter occupants. 13. If the helicopter had been equipped with a crash-resistant fuel system, the potential for thermal injuries to the occupants would have been reduced or eliminated. 14.

[ANALYSIS] Company personnel also stated that it was common practice to slightly increase the collective to lift the helicopter about 3 to 5 feet above the ground and perform a check of systems. Section 1.5.4 discusses the pilot’s failure to perform a hover check during the accident flight. 1.3.5.3 Previous Dual-Hydraulic AS350-Series Helicopter Events Involving Loss of Yaw Control On April 9, 2014, an Airbus Helicopters AS350 B3e crashed into a hospital rooftop in Albuquerque, New Mexico, after departing from the hospital’s heliport. The helicopter was operating under the provisions of 14 CFR Part 91 as a local positioning flight. The pilot and two medical technicians received minor injuries, and the helicopter was substantially damaged.31 The pilot reported that he completed the preflight hydraulic check and did not note anything abnormal 31 The brief report for each event summarized in this accident report can be searched by case number from the NTSB’s Aviation Accident Database web page, and each investigation’s public docket can be accessed from the NTSB’s Accident Dockets web page. All NTSB documents are accessible from the NTSB’s home page (www.ntsb.gov). For more information about this accident, see case number CEN14FA193. NTSB Aircraft Accident Report 18 with the pedal movement. During the takeoff, the pilot expected to make a slight left turn to clear the heliport platform and hospital for departure, but the helicopter kept turning. The pilot attempted to apply right pedal to stop the turn but reported that the pedals felt jammed or locked in the neutral position. The helicopter began spinning to the left. Although the yaw servo hydraulic switch was found in the “ON” position, the switch’s position at takeoff could not be determined because of the lack of an onboard image recorder. The NTSB determined that the probable cause of the accident was “the pilot’s loss of yaw control during takeoff due to the absence of hydraulic boost to the tail rotor pedals for reasons that could not be determined based on the available information.” A finding in the accident was “the lack of a caution indicator to alert the pilot of the lower hydraulic system configuration.” During the investigation of the Albuquerque accident, the NTSB learned about a 2009 event involving a Eurocopter AS350 B3 helicopter with a dual hydraulic system that was operated by the US Customs and Border Protection.

[ANALYSIS] According to Airbus Helicopters and the FAA, the helicopter experienced a loss of yaw control because the yaw servo hydraulic switch was not reset to the “ON” position during the preflight hydraulic check.32 In addition, the NTSB investigated a June 26, 2014, incident involving an Airbus Helicopters AS350 B3 helicopter, N808LF, at the Draughon-Miller Central Texas Regional Airport, Temple, Texas. The helicopter was operated by Air Methods under the provisions of 14 CFR Part 91 as a positioning flight. The pilot reported that, immediately after takeoff, the helicopter started to yaw to the left. He indicated that the pedals were locked in the neutral position and felt jammed and that he attempted to correct the rotation but was unable to do so. The pilot then executed a precautionary hovering autorotation. The pilot and two crewmembers were not injured, and the helicopter received minor damage. A postincident examination of the helicopter found no anomalies with its tail rotor and hydraulic systems, but the yaw servo hydraulic switch was found in its “OFF” position. The NTSB determined that the probable cause of this incident was “the pilot’s failure to reposition the yaw servo hydraulic switch to the ‘on’ position during the pretakeoff hydraulic system check, which resulted in a complete lack of hydraulic boost to the tail rotor system and increased the load required to move the control pedals and led to the pilot’s subsequent inability to manipulate the control pedals and his loss of yaw control.”33 1.3.5.4 Safety Information to Prevent Lack of Hydraulic Boost to Pedals On August 21, 2014, Airbus Helicopters issued Safety Information Notice No. 2776-S-29 to provide information about the preflight hydraulic check for AS350 B3 helicopters with a dual hydraulic system.34 In the notice, Airbus Helicopters stated that it became aware of “at least two 32 The NTSB should have been notified about this event when it occurred but was not notified for undetermined reasons. 33 For more information about this incident, see case number CEN14IA329. 34 At the time of the accident, Airbus Helicopters safety information notices and related documents were sent to the engineering department at Air Methods, and a department staff member reviewed each document to determine the relevant departments within the company that needed to receive the document.

[ANALYSIS] The circumstances leading to the accident are discussed in the sections that follow. 1.5.1 Loss of Tail Rotor Control As previously stated, the helicopter’s increasing left yaw rate was consistent with increasing main rotor torque without a sufficient increase in tail rotor thrust to maintain a steady heading. The three likely causes for insufficient tail rotor thrust are a loss of (1) tail rotor drive, 56 For the partially boosted and nonboosted configurations, pedal control loads increase as the pedal is pushed farther forward. 57 Controllability becomes a major concern when a helicopter is spinning due to the inertial forces that the pilot feels. Specifically, when a helicopter is spinning nose left, the pilot tends to get pushed to the right of the seat, which makes it difficult to maintain steady and coordinated inputs. NTSB Aircraft Accident Report 27 (2) tail rotor effectiveness, and (3) tail rotor control. On the basis of the findings from the investigation of this accident, the NTSB excluded a loss of tail rotor drive and LTE as possible causes for the insufficient tail rotor thrust (see sections 1.3.3.3 and 1.4.2, respectively). Thus, a loss of tail rotor control caused the left yaw at the time of takeoff. The loss of tail rotor control could have resulted from a physical restriction within the tail rotor flight control system, but the lack of evidence (that is, no witness marks) indicating such a restriction within the forward tail rotor flight controls eliminated that possibility.58 The loss of tail rotor control could also have resulted from a disconnect within the tail rotor flight control system, but that possibility was discounted because of evidence indicating that the pilot made a partial right pedal input at takeoff. In addition, no recent maintenance had been performed on the tail rotor flight controls, and no evidence indicated any disconnect within the tail rotor flight controls. Another reason for a loss of tail rotor control is a partial or full loss of hydraulic boost to the pedals. In an AS350 B3e helicopter, a partial or full loss of hydraulic boost to the pedals could be caused by a loss of hydraulic pressure to the lower hydraulic system or a malfunction of the tail rotor hydraulic isolation valve. For either of these conditions, if the tail rotor hydraulic circuit were correctly configured—the yaw servo hydraulic switch reset to the “ON” position and the yaw load compensator accumulator charged—after the yaw servo hydraulic check, the yaw load compensator would provide partial hydraulic assistance to the pedals. Airbus Helicopters calculated that the necessary pedal force to maintain a steady heading with partial hydraulic assistance would be about 30 to 45 pounds. A partial or full loss of hydraulic boost to the pedals could also be caused by an incorrectly configured tail rotor hydraulic circuit (as a result of the yaw servo hydraulic check not being performed correctly).59 If the yaw servo hydraulic switch were not reset to the “ON” position and the yaw load compensator accumulator were depleted, the pilot could experience a lack of hydraulic boost to the pedals.

[ANALYSIS] Second, and more importantly, all maintenance and inspection tasks are subject to human error. As previously discussed, this investigation has identified several weaknesses in the lubrication and inspection procedures that could affect their intended results and compromise safety.272 Further, several Safety Board accident investigations, including the Alaska Airlines flight 261 investigation, have demonstrated that even simple maintenance tasks are sometimes missed or inadequately performed and can have catastrophic results.273 Therefore, the current horizontal stabilizer trim system design remains 271 For a more detailed discussion of the Safety Board’s concerns in this area, see section 2.4.2. 272 For more information about the weaknesses in the lubrication and inspection procedures, see sections 2.3.6.3 and 2.4.3, respectively. 273 For example, following the September 11, 1991, crash of a Continental Express Embraer 120 in Eagle Lake, Texas, the Safety Board concluded that the airline’s maintenance inspection and quality assurance programs failed to detect that the upper row of screws on the leading edge of the left horizontal stabilizer had been removed during maintenance and had not been replaced. The partially secured left horizontal stabilizer leading edge separated in flight, causing a severe nose-down pitchover. The airplane broke up in flight, and all 14 people on board were killed. For more information, see National Transportation Safety Board. Britt Airways, Inc., d/b/a Continental Express Flight 2574, In-flight Structural Breakup, EMB-120RT, N33701, Eagle Lake, Texas, September 11, 1991. NTSB/AAR-92/04. In addition, the Board’s investigation of the July 6, 1996, uncontained engine failure on a Delta Air Lines MD-88 in Pensacola, Florida, determined that a fluorescent inspection process used to detect fatigue cracks during maintenance was susceptible to error because it involved multiple cleaning, processing, and inspection procedures dependent on several individuals and because of a low expectation of finding a crack. Shrapnel from the uncontained engine failure pierced the fuselage and entered the rear cabin. Two passengers were killed, and two others were seriously injured. For more information, see National Transportation Safety Board. Uncontained Engine Failure, Delta Airlines Flight 1288, McDonnell Douglas MD-88, N927DA, Pensacola, Florida, July 6, 1996. NTSB/AAR-98-01.

[ANALYSIS] Shrapnel from the uncontained engine failure pierced the fuselage and entered the rear cabin. Two passengers were killed, and two others were seriously injured. For more information, see National Transportation Safety Board. Uncontained Engine Failure, Delta Airlines Flight 1288, McDonnell Douglas MD-88, N927DA, Pensacola, Florida, July 6, 1996. NTSB/AAR-98-01. Analysis 166 Aircraft Accident Report vulnerable to catastrophic failure if maintenance and inspection tasks are not performed properly. 2.6.3 Elimination of Catastrophic Effects of Acme Nut Thread Loss Through Design The Safety Board concludes that, when a single failure could have catastrophic results and there is a practicable design alternative that could eliminate the catastrophic effects of the failure mode, it is not appropriate to rely solely on maintenance and inspection intervention to prevent the failure from occurring; if a practicable design alternative does not exist, a comprehensive systemic maintenance and inspection process is necessary. In the case of the horizontal stabilizer trim system, such a design would incorporate a reliable, independent means for eliminating, overcoming, or counteracting the catastrophic effects of acme nut thread loss. The Safety Board notes that such a design change would not necessarily need to incorporate dual actuators, or any other form of system redundancy; the design would only need to provide a mechanism for preventing stripped acme nut threads from resulting in unrecoverable movement of the horizontal stabilizer. The Board notes that among the several design concepts listed in AC 25.1309-1A that can be used to avoid catastrophic failure conditions are the following: (1) “designed failure effect limits, including the capability to sustain damage, to limit the safety impact or effects of a failure”; and (2) “designed failure path to control and direct the effects of a failure in a way that limits its safety impact.” The Safety Board concludes that transport-category airplanes should be modified, if practicable, to ensure that horizontal stabilizer trim system failures do not preclude continued safe flight and landing. Therefore, the Safety Board believes that the FAA should conduct a systematic engineering review to (1) identify means to eliminate the catastrophic effects of total acme nut thread failure in the horizontal stabilizer trim system jackscrew assembly in DC-9, MD-80/90, and 717 series airplanes and require, if practicable, that such fail-safe mechanisms be incorporated in the design of all existing and future DC-9, MD-80/90, and 717 series airplanes and their derivatives; (2) evaluate the horizontal stabilizer trim systems of all other transport-category airplanes to identify any designs that have a catastrophic single-point failure mode and, for any such system; (3) identify means to eliminate the catastrophic effects of that single-point failure mode and, if practicable, require that such fail-safe mechanisms be incorporated in the design of all existing and future airplanes that are equipped with such horizontal stabilizer trim systems. Further, the Safety Board is concerned that the FAA certified a horizontal stabilizer trim system that had a single-point catastrophic failure mode.

[ANALYSIS] The captain did not brief the first officer about what to expect or what to do if these configuration changes resulted in excessive flight control pressures or loss of control of the airplane. Further, the captain did not specify that the flaps should be extended at a slower-than-normal rate, which would have been a prudent precaution to minimize the possibility of the configuration change causing abrupt airplane movements that could be difficult to control. Nevertheless, at 1618:17, after the slats and flaps were extended, the captain noted that the airplane was “pretty stable right here.” The captain added that the airspeed needed to decrease to 180 KIAS (the airplane was then at 250 KIAS). Nine seconds later, at 1618:26, the captain ordered retraction of the slats and flaps, and the airspeed began to subsequently increase. It was not clear from the CVR recording why the captain ordered retraction of the slats and flaps and allowed the airspeed to increase nor did the CVR recording indicate any discussion about the possible effects of the slat and flap extension. The Safety Board notes that an airplane with flight control problems should be handled in a slow and methodical manner and that any configuration that would aid a landing should be maintained if possible. On the basis of the captain’s comment, the airplane was stable after the slat and flap extension at 1618:05. This configuration would have aided the approach and landing process. The Safety Board concludes that flight crews dealing with an in-flight control problem should maintain any configuration change that would aid in accomplishing a safe approach and landing, unless that configuration change adversely affects the airplane’s controllability. 2.2.5.4 Activation of the Primary Trim Motor At 1618:49, after the slats and flaps were retracted, the captain stated that he wanted to “get the nose up…and then let the nose fall through and see if we can stab it when it’s unloaded.” The first officer responded, “you mean use this again? I don’t think we should…if it can fly.” These statements suggest that the captain may have been indicating his intention to retry the primary trim system after reducing aerodynamic forces on the horizontal stabilizer. However, after the first officer’s statement at 1619:14, “I think if it’s controllable, we oughta just try to land it,” the captain abandoned his plan and responded, “ok let’s head for LA.” 229 The Safety Board also notes that the captain’s disconnection of the autopilot when he was the pilot not flying was also contrary to standard industry procedures. Normally, the pilot flying would use the autopilot disconnect switch on the control column and then assess the airplane’s controllability.

[ANALYSIS] Therefore, the extremely loud noise recorded on the CVR at 1619:36.6 was likely made as the fairing brackets failed and caused the loss of the tip fairing and structural deformation of the tail under flight loads, resulting in local aerodynamic disturbances.223 Valid airplane performance data do not exist for horizontal stabilizer positions beyond 14º airplane nose down. Although the exact horizontal stabilizer angle required to produce the final pitchover could not be determined, the Safety Board’s airplane performance studies determined that the airplane’s pitch rate during the final dive could only have resulted from a rapid increase to a stabilizer angle significantly beyond 14º airplane nose down.224 In sum, the Safety Board concludes that the cause of the final dive was the low-cycle fatigue fracture of the torque tube, followed by the failure of the vertical stabilizer tip fairing brackets, which allowed the horizontal stabilizer leading edge to move upward significantly beyond what is permitted by a normally operating jackscrew assembly. The resulting upward movement of the horizontal stabilizer leading edge created an excessive upward aerodynamic tail load, which caused an uncontrollable downward pitching of the airplane from which recovery was not possible. 2.2.5 Flight Crew Decision-Making 2.2.5.1 Decision to Continue Flying Rather than Return to PVR Safety Board investigators considered several reasons that might explain the captain’s decision not to return immediately to PVR after he experienced problems with the horizontal stabilizer trim system during the climbout from PVR. Neither the Alaska Airlines MD-80 Quick Reference Handbook (QRH) Stabilizer Inoperative checklist nor the company’s QRH Runaway Stabilizer emergency checklist required landing at the nearest suitable airport if corrective actions were not successful. These checklist procedures were the only stabilizer-related checklist procedures contained in the QRH, and the flight crew most likely followed these checklist procedures in their initial attempts to correct the airplane’s jammed stabilizer. 222 The Safety Board notes that the fairing brackets were not intended to carry the load of the horizontal stabilizer. 223 As discussed in section 1.11.1.1, sound tests indicated that neither this sound nor any of the other sounds recorded by the accident airplane’s CVR were similar to sounds associated with the failure of the torque tube. The investigation could not determine exactly what caused the sound. 224 Structural analysis indicated that the horizontal stabilizer must travel to a position of greater than 15º airplane nose down to allow the acme screw to completely exit the top of the acme nut, as it did during the final dive. Symmetrical structural damage was found on the tail, indicating that the horizontal stabilizer’s hinge fittings contacted the vertical stabilizer’s rear spar, which would occur at a horizontal stabilizer position of 16.2º airplane nose down. Analysis 136 Aircraft Accident Report The airplane’s takeoff weight of 136,513 pounds was well below the takeoff and climb limits for the departure runway, but it exceeded the airplane’s maximum landing weight of 130,000 pounds.

[CONCLUSIONS > FINDINGS] The pilots recognized that the longitudinal trim control system was jammed, but neither they nor the Alaska Airlines maintenance personnel could determine the cause of the jam. 10. The worn threads inside the horizontal stabilizer acme nut were incrementally sheared off by the acme screw and were completely sheared off during the accident flight. As the airplane passed through 23,400 feet, the acme screw and nut jammed, preventing further movement of the horizontal stabilizer until the initial dive. Conclusions 177 Aircraft Accident Report 11. The accident airplane’s initial dive from 31,050 feet began when the jam between the acme screw and nut was overcome as a result of operation of the primary trim motor. Release of the jam allowed the acme screw to pull up through the acme nut, causing the horizontal stabilizer leading edge to move upward, thus causing the airplane to pitch rapidly downward. 12. The acme screw did not completely separate from the acme nut during the initial dive because the screw’s lower mechanical stop was restrained by the lower surface of the acme nut until just before the second and final dive about 10 minutes later. 13. The cause of the final dive was the low-cycle fatigue fracture of the torque tube, followed by the failure of the vertical stabilizer tip fairing brackets, which allowed the horizontal stabilizer leading edge to move upward significantly beyond what is permitted by a normally operating jackscrew assembly. The resulting upward movement of the horizontal stabilizer leading edge created an excessive upward aerodynamic tail load, which caused an uncontrollable downward pitching of the airplane from which recovery was not possible. 14. In light of the absence of a checklist requirement to land as soon as possible and the circumstances confronting the flight crew, the flight crew’s decision not to return to Lic Gustavo Diaz Ordaz International Airport, Puerto Vallarta, Mexico, immediately after recognizing the horizontal stabilizer trim system malfunction was understandable. 15. The flight crew’s decision to divert the flight to Los Angeles International Airport, Los Angeles, California, rather than continue to San Francisco International Airport, San Francisco, California, as originally planned was prudent and appropriate. 16. Alaska Airlines dispatch personnel appear to have attempted to influence the flight crew to continue to San Francisco International Airport, San Francisco, California, instead of diverting to Los Angeles International Airport, Los Angeles, California. 17. The flight crew’s use of the autopilot while the horizontal stabilizer was jammed was not appropriate. 18. Flight crews dealing with an in-flight control problem should maintain any configuration change that would aid in accomplishing a safe approach and landing, unless that configuration change adversely affects the airplane’s controllability. 19. Without clearer guidance to flight crews regarding which actions are appropriate and which are inappropriate in the event of an inoperative or malfunctioning flight control system, pilots may experiment with improvised troubleshooting measures that could inadvertently worsen the condition of a controllable airplane. 20. The acme nut threads on the accident airplane’s horizontal stabilizer jackscrew assembly wore at an excessive rate.

[CONCLUSIONS] The design of the Douglas DC-9, McDonnell Douglas MD-80/90, and Boeing 717 horizontal stabilizer jackscrew assembly did not account for the loss of the acme nut threads as a catastrophic single-point failure mode. The absence of a fail-safe mechanism to prevent the catastrophic effects of total acme nut thread loss contributed to the Alaska Airlines flight 261 accident. Conclusions 180 Aircraft Accident Report 41. When a single failure could have catastrophic results and there is a practicable design alternative that could eliminate the catastrophic effects of the failure mode, it is not appropriate to rely solely on maintenance and inspection intervention to prevent the failure from occurring; if a practicable design alternative does not exist, a comprehensive systemic maintenance and inspection process is necessary. 42. Transport-category airplanes should be modified, if practicable, to ensure that horizontal stabilizer trim system failures do not preclude continued safe flight and landing. 43. Catastrophic single-point failure modes should be prohibited in the design of all future airplanes with horizontal stabilizer trim systems, regardless of whether any element of that system is considered structure rather than system or is otherwise considered exempt from certification standards for systems. 44. The certification requirements applicable to transport-category airplanes should fully consider and address the consequences of failures resulting from wear. 45. At the time of the flight 261 accident, Alaska Airlines’ maintenance program had widespread systemic deficiencies. 46. The Federal Aviation Administration (FAA) did not fulfill its responsibility to properly oversee the maintenance operations at Alaska Airlines, and at the time of the Alaska Airlines flight 261 accident, FAA surveillance of Alaska Airlines had been deficient for at least several years. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was a loss of airplane pitch control resulting from the in-flight failure of the horizontal stabilizer trim system jackscrew assembly’s acme nut threads. The thread failure was caused by excessive wear resulting from Alaska Airlines’ insufficient lubrication of the jackscrew assembly. Contributing to the accident were Alaska Airlines’ extended lubrication interval and the Federal Aviation Administration’s (FAA) approval of that extension, which increased the likelihood that a missed or inadequate lubrication would result in excessive wear of the acme nut threads, and Alaska Airlines’ extended end play check interval and the FAA’s approval of that extension, which allowed the excessive wear of the acme nut threads to progress to failure without the opportunity for detection. Also contributing to the accident was the absence on the McDonnell Douglas MD-80 of a fail-safe mechanism to prevent the catastrophic effects of total acme nut thread loss. Recommendations 181 Aircraft Accident Report 4. Recommendations 4.1 New Recommendations As a result of the investigation of the Alaska Airlines flight 261 accident, the National Transportation Safety Board makes the following recommendations to the Federal Aviation Administration: Issue a flight standards information bulletin directing air carriers to instruct pilots that in the event of an inoperative or malfunctioning flight control system, if the airplane is controllable they should complete only the applicable checklist procedures and should not attempt any corrective actions beyond those specified.

[ANALYSIS] Structural examinations determined that this additional movement of the horizontal stabilizer could only result from a fracture of the acme screw torque tube. Metallurgical analysis of the recovered section of the fractured torque tube showed that the final separation initiated from an area of low-cycle fatigue cracking. On the basis of an evaluation of the static loading conditions and performance data, the Safety Board determined that the torque tube was capable of sustaining a single application of the maximum load applied to the torque tube between the initial and final dives.220 However, low-cycle fatigue tests and analysis, performance data, and FEA established that the magnitude and frequency of the loads on the torque tube after complete shearing of the acme nut threads, in combination with an offset loading condition,221 would be sufficient to initiate and propagate low-cycle fatigue cracking in the torque tube. The low-cycle fatigue cracking severely reduced the load capability of the torque tube during the 10-minute period between the initial and second and final dive, causing it to fracture at loads considerably less than it was capable of sustaining if it did not have fatigue damage. At 1619:29, the captain ordered redeployment of the slats and flaps. At 1619:35, the flaps were transitioning from 7º to 11º. At 1619:36.6, the CVR recorded the sound of an “extremely loud noise.” Immediately thereafter, the airplane began its final dive. At the time of this pitchover, radar detected several small primary returns, consistent with parts of the vertical stabilizer’s tip fairing being torn from the airplane as the fairing brackets 218 According to the Safety Board’s sound spectrum study, the “faint thump” noted on the CVR transcript at 1619:21was actually a series of thumps. 219 After the sound of the faint thump, the elevator deflections increased from -9º to -13º, indicating the pilots were making additional airplane-nose-up elevator inputs to maintain level flight with the increased airplane-nose-down horizontal stabilizer angle. 220 For more information about these loads, see section 1.16.5 221 The torque tube was subjected to a combination of bending and tension as a result of an offset loading condition created in the torque tube because the stop lugs on the lower mechanical stop and the acme screw prevented application of equal load around the axis of the tube. Analysis 135 Aircraft Accident Report broke,222 which would have allowed the horizontal stabilizer to move well beyond the 3.6º airplane-nose-down position it was being held at by the brackets. Therefore, the extremely loud noise recorded on the CVR at 1619:36.6 was likely made as the fairing brackets failed and caused the loss of the tip fairing and structural deformation of the tail under flight loads, resulting in local aerodynamic disturbances.223 Valid airplane performance data do not exist for horizontal stabilizer positions beyond 14º airplane nose down.

[ANALYSIS] The 2 flight crewmembers, 1 cabin crewmember and 11 passengers aboard the airplane were fatally injured. The National Transportation Safety Board determines that the probable cause of this accident was the failure of Continental Express maintenance and inspection personnel to adhere to proper maintenance and quality assurance procedures for the airplane's horizontal stabilizer deice boots that led to the sudden in-flight loss of the partially secured left horizontal stabilizer leading edge and the immediate severe nose-down pitchover and breakup of the airplane. Contributing to the cause of the accident was the failure of the Continental Express management to ensure compliance with the approved maintenance procedures, and the failure of FAA surveillance to detect and verify compliance with approved procedures. - The issues in this investigation focused on: 1. The responsibilities of the Federal Aviation Administration and aircraft manufacturers and operators to determine the critical items and inspection levels of aircraft systems. 2. The procedures for relaying and standardizing maintenance shift turnover information. As a result of this investigation, the Safety Board issued safety recommendations to the Federal Aviation Administration on the feasibility of developing a means to advise flightcrews of recent maintenance work on aircraft and the need for reviewing regulations, policies and practices for establishing required inspection items with a view toward developing more specific identification of such items. Also, as a result of this investigation, on February 28, 1992, the Safety Board issued safety recommendations to the Federal Aviation Administration that would enhance both flight standards surveillance of Continental Express and flight standards Program Guidelines, including the National Aviation Safety Inspection Program. I z. NATIONAL TRANSPORTATION SAFETY BOARD WASHINGTON, D.C. 20594 AIRCRAFT ACCIDENT REPORT BRITT AIRWAYS, INC., d/b/a CONTINENTAL EXPRESS FLIGHT 2574 - IN-FLIGHT STRUCTURAL BREAKUP EMB-120RT, N33701 EAGLE LAKE, TEXAS SEPTEMBER 11, 1991 1. FACTUAL INFORMATION

[ANALYSIS] Further, the lack of fire damage on the left engine suggests that this engine separated early in the breakup sequence when the left wing failed. The - failure of the left wing released fuel that probably led to the in-flight fire. The passenger seat that was ejected from the cabin at ground impact suggests that the fire did not progress into the cabin area before impact. This I conclusion is supported by the absence of soot deposits in the respiratory tract of the occupants, and the absence of elevated carboxyhemoglobin in the tissues of the occupants. . The FDR data and examination of the wreckage revealed that the flight control systems, engines, and propellers were operating normally before the extreme attitude changes of the airplane. Consequently, engine and propeller malfunctions were not a factor in the accident. . The Safety Board's analysis of this accident included an examination of the circumstances that led to the loss of the left stabilizer leading edge, including: flightcrew performance related to the accident; the maintenance and inspection conducted by Continental Express the night before the accident; the management of the Continental Express maintenance department; the FAA 34 approval and oversight of the Continental Express maintenance program; and the procedures for establishing RIIs by the aircraft manufacturer, the airline, and the FAA. The Safety Board’ s analysis also examined the aerodynamic and . structural failure aspects related to the dynamics of the airplane after it lost the left stabilizer leading edge. 2.2 Aerodynamic and Structural Failure Aspects The Safety Board believes that the airplane experienced the following sequence of events during the final moments of flight. The airplane was descending at 260 KIAS, which was well within its operating envelope, the wings were level, both engines were operating normally, and the pitch attitude was 10 degrees nose down. As the airplane descended through 11,500 feet, the leading edge of the left horizontal stabilizer separated from the airframe. The left horizontal stabilizer leading edge was the first piece of wreckage found along the wreckage path, preceding the next piece by almost 1/2 mile. This indicates that it _ was the first piece to separate from the airplane. The loss of the leading edge exposed the front spar of the left side of the horizontal stabilizer to the airstream, and an aerodynamic stall occurred that greatly reduced the downforce produced by the horizontal stabilizer. The reduction in downforce created a large nose-down pitching moment, and the airplane pitched down immediately. A peak load factor of approximately -5 g was reached at the end of only 1 second. The airframe remained intact (minus the leading edge), and the load factor fluctuated around -2 g, for approximately 6 1/2 seconds. The. airplane pitch attitude decreased to 68 degrees nose down, airplane heading moved 20 degrees nose left, and a 15 degree right roll attitude was reached at the end of this period. The airplane's altitude was 9,500 feet, and it was flying at an airspeed of 280 KIAS.

[ANALYSIS] The. airplane pitch attitude decreased to 68 degrees nose down, airplane heading moved 20 degrees nose left, and a 15 degree right roll attitude was reached at the end of this period. The airplane's altitude was 9,500 feet, and it was flying at an airspeed of 280 KIAS. A second peak in negative load factor was then experienced, and the Safety Board believes that the left wing failed and the right wing tip detached at _ this point. The airplane then rolled to the right at a roll rate exceeding 160 degrees per second. The Safety Board believes that the lift produced by the intact right wing produced the extreme roll. The high airspeed and roll rate created large airloads on the airplane's structure. The Safety Board believes: that excessive — airloads induced by the high airspeeds and/or roll rate caused the horizontal stabilizer and left engine to separate from the airframe. The airplane then entered a 35 spin to the right, fell uncontrollably toward impact, its pitch attitude oscillating between approximately -40 degrees and +40 degrees. To recreate this sequence, the Safety Board relied on the substantial amount of evidence obtained from the wreckage, CVR, and FDR. Flight dynamics and structural simulations by Embraer provided additional data for use in the investigation. The Safety Board analyzed the airloads that were applied to the partially secured leading edge on the accident airplane. The atmosphere was calm; therefore, gust loads were probably not a factor in the separation of the leading edge. Aerodynamic lift and drag both produce loads on the horizontal stabilizer structure. In general, aerodynamic loads are significantly greater at higher airspeeds since the dynamic pressure of the airstream varies directly with the square of airplane velocity. Aerodynamic drag exerts a force on the airplane that is opposite to the direction of motion and parallel to the relative wind. Therefore, aerodynamic drag created an aft load on the horizontal stabilizer structure of the accident airplane. This force compressed the partially secured leading edge against the front spar of the stabilizer structure and helped to keep the leading edge in place. However, aerodynamic lift is also an important factor in the determination of airloads acting on the stabilizer. The horizontal stabilizer in this case provides negative, or downward, lift to balance the pitching moment of the wings, engines, and fuselage. Airplane nose pitch attitude is controlled up or down by deflecting the elevator attached to . the rear of the horizontal stabilizer. The lift force required at the horizontal stabilizer to establish trimmed flight is a function of many factors, such as the center of gravity, engine thrust, airspeed, and airplane configuration. The airplane is described as being "trimmed" in pitch if the sum of the pitching moments created by these factors is equal to zero. Calculations that defined the horizontal stabilizer lift required (downward) for the two postmaintenance flights on N33701 showed that the peak download occurred at the time of the accident flight upset.

[ANALYSIS] The FAA responded to these two safety recommendations in a letter dated May 15, 1992, indicating its agreement with the needs expressed in the recommendations. The Safety Board's response to this letter, as well as to other FAA letters that address open safety recommendations about FAA surveillance of air Carrier operations and maintenance practices, is attached as Appendix I. 3.1 Findings 1. All crewmembers and air traffic controllers were properly certified to perform their duties. 2. There was no evidence of flightcrew activities during the preflight inspection or. during the accident flight that were causal to this accident. 3. There was no evidence of air traffic controller activity that was - causal to this accident . . 4. I Weather was not a factor in the accident. 5. There was no evidence of engine or flight control! malfunctions. 6. The accident was precipitated by the loss of the left horizontal ~ - stabilizer leading edge when the airplane was in a descent 12 knots below its maximum safe operating speed, within its operating envelope. . 7. The airplane pitched severely nose down upon the loss of the ~ left horizontal stabilizer leading edge, and the wings stalled - negatively. ] ) 8. The violent motion of the airplane and the extreme airloads that ‘ resulted from the loss ef the left horizontal stabilizer leading edge caused the airplane to break up in flight. . 9. Anin flight fire occurred during the structural breakup. 10. The left horizontal stabilizer leading edge separated from the airplane because the upper row of screw fasteners (47) was not in place. The airloads during the descent caused the surface to bend downward and separate. _ The upper row of fasteners for the left horizontal stabilizer UL 48 3. CONCLUSIONS leading edge had been removed during scheduled maintenance 12, 13. 14. 15. 16. 17. 18. 19, 49 the night before the accident, and a breakdown in procedures failed to detect that the work was incomplete. The Continental Express FAA-approved General Maintenance Manual (GMM) contained adequate procedures for _ Maintenance and quality control. There was a lack of compliance with the GMM procedures by the mechanics, inspectors, and supervisors responsible for ensuring the airworthiness of N33701 the night before the I accident. The lack of compliance with the GMM procedures by the Continental Express maintenance department led to the return of an unairworthy airplane to scheduled passenger service.

[ANALYSIS] This information is consistent with the FDR data and physical evidence that the empennage did not fail until after the failure of the left wing. Witnesses reported seeing the airplane in a left spin prior to ground impact. Although the FDR data show the airplane in a right spin following wing failure, the recording ends about 13 seconds before impact. In summary, ‘the Safety Board concludes that the FDR data, engineering simulation, and examination of the wreckage confirm that the accident sequence was initiated by the loss of the left leading edge of the horizontal stabilizer. 37 2.3. _ Flightcrew Preflight Performance The Safety Board found no evidence that the two pilots were informed of the work that had been performed on the horizontal stabilizer the night before the accident. Of course, if the pilots. had wanted to review the maintenance records for the airplane, the records would undoubtedly have been made available to them. However, there was no indication of any work on the stabilizer leading edges in the pilot's airplane log book, and no indication has been found that the flightcrew was informed of any of this work, even though the work was on a critical assembly of the airplane--the horizontal stabilizer leading edges. . The Safety Board is aware that the work performed on the horizontal stabilizers was considered scheduled maintenance and was not normally noted in the pilot's airplane log book. Further, there are no regulatory provisions for pilots to be made aware of routine maintenance work, regardless of its complexity. However, the Safety Board believes that a study should be undertaken on the feasibility of developing a means to advise flightcrews about recent maintenance actions, both routine and nonroutine, of the airplanes they are about to fly; so that they have the opportunity to be alert to discrepancies during preflight inspections and possibly to make an additional inspection of critical items, such as RIs, that may affect the safety of flight. In this case, if the flightcrew had been informed of the previous night's work on the airplane, they might have, with the advantage of moming daylight, lent a crucial hand in checking the work. The top of the horizontal stabilizer on the airplane's "T-tail" is about 20 feet above the ground. Therefore, the flightcrew could not have seen the area of the missing screws on top of the leading edge/deice boot during their normal preflight inspection. However, if they had been informed of the maintenance, they might have discussed the work with maintenance personnel and requested them to conduct a visual inspection of the stabilizer's upper surface. Because the - flightcrew was unaware of the previous night's work on the airplane, the possibility of having another set of eyes observe the work was eliminated. The Safety Board believes that the FAA should require airlines to establish procedures to inform pilots of all significant maintenance on airplanes before flight. Such information would allow pilots to be more alert to potential unsafe conditions when they conduct preflight inspections.

[ANALYSIS] Analysis 273 Aircraft Accident Report upsets of USAir flight 427, United flight 585, and Eastwind flight 517 were most likely caused by the movement of the rudder surfaces to their blowdown limits in a direction opposite to that commanded by the pilots. The rudder surfaces most likely moved as a result of jams of the secondary slides to the servo valve housings offset from their neutral position and overtravel of the primary slides. In addition to this reversal potential, the Safety Board’s investigation revealed two other potential failure mechanisms353 within the 737 rudder control system that could result in a deflection to the rudder’s blowdown limit. One of these potential failure mechanisms is a physical jam in the rudder system input linkage (between the PCU’s input crank and body stop), preventing the main rudder PCU control valve from closing; the other is a jam of the primary to the secondary slide of the main rudder PCU servo valve combined with a jam of the secondary slide to the servo valve housing at positions other than neutral (known as a dual jam). These failure mechanisms probably did not play a role in the USAir flight 427, United flight 585, and Eastwind 517 upsets.354 Nonetheless, the failure mechanisms are cause for concern because they further illustrate the vulnerability of the 737 rudder system to jams that could produce rudder deflections and result in catastrophic consequences. 2.6 Adequacy of the Boeing 737 Rudder System Design Boeing has recently made significant design changes in the 737 rudder system, especially on the 737-NG. (The design changes on the NG series airplanes include a redesigned main rudder PCU servo valve in which the hydraulic fluid ports are spread, thus eliminating the reversal mechanism identified in the thermal tests; a redesigned yaw damper system; a hydraulic pressure limiter; a rudder input force transducer; and a new standby rudder PCU input bearing.) The 737-100 through -500 series airplanes are being retrofitted with the redesigned servo valve and a hydraulic pressure reducer designed to limit the extent to which the airplanes would be vulnerable to the rudder overpowering the roll authority of the ailerons and spoilers. As a result of ADs issued by the FAA, the redesigned main rudder PCU servo valve should eliminate the possibility of a rudder reversal from the specific circumstances 353 A third potential failure mechanism—a jam of the primary to the secondary slide with overtravel of the secondary slide—was identified as a result of testing after the July 1992 United Airlines rudder anomaly that occurred during a ground check.

[CONCLUSIONS] Conclusions 293 Aircraft Accident Report 10. Analysis of the human performance data (including operational factors) does not support a scenario in which the flight crew of USAir flight 427 applied and held a full left rudder input until ground impact more than 20 seconds later. 11. Analysis of the cockpit voice recorder, National Transportation Safety Board computer simulation, and human performance data (including operational factors) from the USAir flight 427 accident shows that they are consistent with a rudder reversal most likely caused by a jam of the main rudder power control unit servo valve secondary slide to the servo valve housing offset from its neutral position and overtravel of the primary slide. 12. The flight crew of USAir flight 427 recognized the initial upset in a timely manner and took immediate action to attempt a recovery but did not successfully regain control of the airplane. 13. The flight crew of USAir flight 427 could not be expected to have assessed the flight control problem and then devised and executed the appropriate recovery procedure for a rudder reversal under the circumstances of the flight. 14. It is very unlikely that the loss of control in the United flight 585 accident was the result of an encounter with a mountain rotor. 15. Analysis of the cockpit voice recorder, National Transportation Safety Board computer simulation, and human performance data (including operational factors) from the United Airlines flight 585 accident shows that they are consistent with a rudder reversal most likely caused by a jam of the main rudder power control unit servo valve secondary slide to the servo valve housing offset from its neutral position and overtravel of the primary slide. 16. The flight crew of United flight 585 recognized the initial upset in a timely manner and took immediate action to attempt a recovery but did not successfully regain control of the airplane. 17. The flight crew of United flight 585 could not be expected to have assessed the flight control problem and then devised and executed the appropriate recovery procedure for a rudder reversal under the circumstances of the flight. 18. Training and piloting techniques developed as a result of the USAir flight 427 accident show that it is possible to counteract an uncommanded deflection of the rudder in most regions of the flight envelope; such training was not yet developed and available to the flight crews of USAir flight 427 or United flight 585. 19. During the Eastwind flight 517 incident, the rudder reversed, moving to its right blowdown limit when the captain commanded left rudder, consistent with a jam of the main rudder power control unit servo valve secondary slide to the servo valve housing offset from its neutral position and overtravel of the primary slide. Conclusions 294 Aircraft Accident Report 20.

[ANALYSIS] The airplanes’ flight attitudes never exceeded reasonable levels, and the airplanes never diverged from controlled flight. The control inputs, bank angles, and headings all converged on stable values, and all of the flights landed safely. During the USAir flight 427 upset, the airplane’s increasingly extreme left bank attitude would have provided the pilots with a consistent and powerful cue to remove any left control inputs they may have applied. Further, the Safety Board’s review of available data from previous accidents and incidents obtained from Boeing, the Board’s database, and accident investigation authorities worldwide indicated that momentary, incorrect rudder applications by air carrier pilots have occasionally occurred in response to an unexpected anomaly during a critical phase of flight. However, those events often occurred in conditions of reduced external visual cues or during abrupt, rapid aircraft movements and accelerations. Some pilots reported being startled by the in-flight upset, but no case was found in which a pilot responded to an in-flight upset involving a sustained yaw or roll by continuing to hold extreme rudder input in a direction opposite to that required to recover the airplane. To further evaluate the possibility of a sustained, inappropriate rudder input, the Safety Board examined numerous possible explanations for the flight crew to have applied and sustained a full left rudder input until a loss of control occurred. These possibilities included pilot incapacitation, deliberate pilot action, disorientation, and unintended rudder pedal activation. The Safety Board reviewed documentation from two incidents in which pilot (specifically, first officer) incapacitation adversely affected the controllability of a 737. In both cases, the incapacitation occurred suddenly; the first officers stiffened and applied pressure to a rudder pedal, resulting in a large rudder deflection as the airplanes descended during the approach to their destination airports. Although both captains reported that they were startled by the unexpected event, both responded appropriately and were capable of compensating for inputs made by the incapacitated first officers. Neither of these cases resulted in a significant loss of control. Further, a review of the available data from previous accidents and incidents revealed no evidence of pilot incapacitation that had resulted in a loss of control in other air carrier airplanes. In the case of USAir flight 427, no evidence indicated that pilot incapacitation was involved in the accident sequence because • neither pilot had a medical history that indicated a risk of incapacitation, • CVR evidence indicated that both pilots were alert and appropriately responsive during the accident sequence, and • CVR evidence indicated that neither pilot was alarmed by the behavior of the pilot or that any medical emergency was occurring. The Safety Board considered whether either pilot deliberately applied an incorrect rudder input. However, the remarks and sounds recorded by the CVR during the initial Analysis 254 Aircraft Accident Report upset and loss of control indicated that both pilots were surprised by the event and did not understand its nature or cause. Further, the Safety Board carefully examined aspects of the pilots’ personal and professional lives and found both pilots to be stable. Moreover, analysis of the communications and sounds recorded by the CVR indicated that the pilots expended extraordinary effort in their attempts to recover from the upset throughout the accident sequence.

[ANALYSIS] These factors combined to produce a flight situation and control problems that the pilots of USAir flight 427 had never before encountered in flight or training, including during stickshaker/stall recovery training. With this series of problems in the course of a few seconds, it is understandable that the crew was no longer responding in a manner that might have allowed recovery.328 During postaccident simulator tests,329 test subjects were able to recover from the USAir flight 427 upset, or at least stabilize the roll to the point at which a continued loss of control would most likely not have occurred, when they applied a specific recovery technique (full right control wheel maintained throughout the duration of the event and forward control column pressure sufficient to reduce G load and maintain a speed above the crossover airspeed) promptly when the event began. However, unlike the pilots of USAir flight 427, the simulator test subjects were aware of the circumstances of the accident, prepared for and expecting the upset event as it occurred, and coached through the recovery procedure. When the simulator test subjects varied their responses from the specific techniques that they were told to apply (for example, when they modified their control wheel input in anticipation of the simulator’s responses to their inputs), a successful recovery from the upset event became much less likely. Further, when the simulator test subjects tried to maintain altitude at the outset of the event, the simulator’s speed decreased below the crossover airspeed, and recovery became unlikely. Therefore, although it was possible to recover from the upset event during its early stages, such a recovery would have required the pilots to immediately abandon their normal pitch control criterion (maintaining altitude) and hold full control wheel inputs against the roll. These actions may be successful with prior awareness of the effects of a 327 See, for example, National Transportation Safety Board. 1997. Uncontrolled Flight Into Terrain, ABX Air Inc. (Airborne Express), Douglas DC-8-63, N827AX, Narrows, Virginia, December 22, 1996. Aircraft Accident Report NTSB/AAR-97/05. Washington, DC. 328 No reliable aerodynamic model exists for the 737’s flight characteristics in a stall; consequently, the Safety Board could not evaluate the possibility of recovery after activation of the stickshaker.

[ANALYSIS] Therefore, no direct comparison was possible between the sounds heard on the flight test airplane and the accident airplane. 307 The Safety Board’s database contains information regarding aviation accidents beginning in 1962. Analysis 245 Aircraft Accident Report Safety Board concludes that, although USAir flight 427 encountered turbulence from Delta flight 1083’s wake vortices, the wake vortex encounter alone would not have caused the continued heading change that occurred after 1903:00. Boeing and Safety Board flight and computer simulations (discussed in section 1.16.6.1) have demonstrated, however, that the heading change rates recorded by the FDR after 1903:00 were consistent with the rudder being deflected to its left aerodynamic blowdown limit. Accordingly, the Safety Board concludes that, about 1903:00, USAir flight 427’s rudder deflected rapidly to the left and reached its left aerodynamic blowdown limit shortly thereafter. This movement of the airplane’s rudder could only have been caused by a flight crew action or a mechanical rudder system anomaly. The potential for such a mechanical rudder anomaly was demonstrated during postaccident tests in which the secondary slide was intentionally jammed (pinned) to the servo valve housing and a rapid input was applied in a direction that would oppose the jam. These tests showed that the primary slide could overtravel,308 resulting in hydraulic fluid porting in such a way that the rudder moves to its aerodynamic blowdown position in the direction opposite to the rudder input (rudder reversal).309 Further, during the most severe postaccident thermal tests (a temperature difference of about 180° between the heated hydraulic fluid and the servo valve housing of the USAir flight 427 main rudder PCU), the secondary slide jammed to the servo valve housing, and hydraulic fluid flow data indicated that a momentary reversal of the rudder occurred during this jam. Although the USAir flight 427 servo valve jammed repeatedly during these extreme thermal tests, the new-production servo valve also subjected to these tests never jammed. Examination of the internal measurements of both servo valves indicated that the USAir flight 427 servo valve had significantly tighter diametrical clearances between the secondary slide and the servo valve housing than the newproduction servo valve. The Safety Board considers it likely that the USAir flight 427 servo valve was more susceptible to a jam because of its tighter clearances. Although the USAir flight 427 main rudder PCU servo valve had been subjected to impact forces from the accident and extensive postaccident testing (including repeated thermal jams), internal examination of the servo valve revealed no evidence of physical marks that would indicate that a jam had existed. Further, the servo valve slides still moved freely, and the servo valve was still capable of successfully completing Parker Hannifin’s acceptance test procedure functional tests.

[ANALYSIS] Pilots can apply rudder inputs during the takeoff or landing ground roll as they use the rudder pedals for nosewheel steering; these inputs can occur at low altitude with a loss of engine power or during a turbulence encounter. Any malfunction resulting in uncommanded rudder motion during an engine failure or in turbulence at low altitude may be catastrophic because of the limited time, altitude, and roll control authority to regain control of the airplane. The Board is also concerned that the limited period of vulnerability to rudder malfunction is based on the assumption that a pilot will perform perfectly and that all airplane systems will perform normally. For example, according to Boeing’s fault tree analysis for the 737-NG, the combination of a jammed servo valve with a loss of engine power during takeoff would be catastrophic only during a 7-second window from V1 through liftoff, at which point roll controls could be used to help control the airplane in the event of a loss of engine power. However, Boeing’s analyses apparently assumed that a pilot would always react immediately and correctly and that the hydraulic pressure limiter would not fail. Such assumptions may not be fully warranted. The Safety Board recognizes that the potential for the specific rudder malfunction that was most likely involved in the accidents of USAir flight 427 and United flight 585 and the incident involving Eastwind flight 517 appears to be have been eliminated by the redesigned servo valve. However, the Board remains concerned that other rudder system malfunctions might potentially lead to rudder reversal or hardover conditions in the 737. The 737 has a history of rudder system-related anomalies, including numerous instances of jamming. Examples of jamming events355 include the following: • a shotpeen ball lodged in a servo valve, causing the rudder to move full right on landing; • shotpeen balls found in a servo valve during a PCU examination; 355 See section 1.18.1.1 for more details about these events. Analysis 275 Aircraft Accident Report • contamination of a PCU by metal particles, causing the rudder pedals to jam during taxi; • internal PCU contamination and worn seals, causing the rudder to lock up on approach; • internal PCU corrosion found during a PCU overhaul; • a loose servo valve retaining nut, causing rudder binding during a flight check and reduced rates, stall, and reversals during testing; • corrosion of a standby rudder PCU, causing full left rudder deflection during taxi; • installation of an incorrect servo valve spring guide, allowing for rudder reversal when the primary slide was jammed to the secondary slide and a rapid rudder input was applied; • fluid contamination of a yaw damper coupler, causing rapid full yaw damper inputs and a severe oscillatory roll; • installation of an incorrect fastener in the summing lever bearing, resulting in a cracked bearing race; and • a jammed or restricted input arm, causing full rudder to move to its full deflection.

[CONCLUSIONS] During the Eastwind flight 517 incident, the rudder reversed, moving to its right blowdown limit when the captain commanded left rudder, consistent with a jam of the main rudder power control unit servo valve secondary slide to the servo valve housing offset from its neutral position and overtravel of the primary slide. Conclusions 294 Aircraft Accident Report 20. It is possible that, in the main rudder power control units from the USAir flight 427, United flight 585, and Eastwind flight 517 airplanes (as a result of some combination of tight clearances within the servo valve, thermal effects, particulate matter in the hydraulic fluid, or other unknown factors), the servo valve secondary slide could jam to the servo valve housing at a position offset from its neutral position without leaving any obvious physical evidence and that, combined with a rudder pedal input, could have caused the rudder to move opposite to the direction commanded by a rudder pedal input. 21. The upsets of USAir flight 427, United flight 585, and Eastwind flight 517 were most likely caused by the movement of the rudder surfaces to their blowdown limits in a direction opposite to that commanded by the pilots. The rudder surfaces most likely moved as a result of jams of the secondary slides to the servo valve housings offset from their neutral position and overtravel of the primary slides. 22. When completed, the rudder system design changes to the Boeing 737 should preclude the rudder reversal failure mode that most likely occurred in the USAir flight 427 and United flight 585 accidents and the Eastwind flight 517 incident. 23. Rudder design changes to Boeing 737-next-generation series airplanes and the changes currently being retrofitted on the remainder of the Boeing 737 fleet do not eliminate the possibility of other potential failure modes and malfunctions in the Boeing 737 rudder system that could lead to a loss of control. 24. The dual-concentric servo valve used in all Boeing 737 main rudder power control units is not reliably redundant. 25. A reliably redundant rudder actuation system is needed for the Boeing 737, despite significant improvements made in the system’s design. 26. The results of this investigation have disclosed that the Boeing 737 rudder system design certificated by the Federal Aviation Administration is not reliably redundant. 27. Transport-category airplanes should be shown to be capable of continued safe flight and landing after a jammed flight control in any position unless the jam can be shown to be extremely improbable. 28. Pilots would be more likely to recover successfully from an uncommanded rudder reversal if they were provided the necessary knowledge, procedures, and training to counter such an event. 29. A neutral rudder pedal position is not a valid indicator that a rudder reversal in the Boeing 737 has been relieved. 30. The training being provided to many Boeing 737 flight crews on the procedures for recovering from a jammed or restricted rudder (including a rudder reversal) is inadequate. Conclusions 295 Aircraft Accident Report 31.

[ANALYSIS] As a result of ADs issued by the FAA, the redesigned main rudder PCU servo valve should eliminate the possibility of a rudder reversal from the specific circumstances 353 A third potential failure mechanism—a jam of the primary to the secondary slide with overtravel of the secondary slide—was identified as a result of testing after the July 1992 United Airlines rudder anomaly that occurred during a ground check. Although the testing determined that this mechanism could cause a rudder reversal, Boeing indicated that subsequent design changes in the servo valve eliminated this possibility. 354 The Safety Board’s postaccident examination of the USAir flight 427 rudder components revealed that the rudder system feedback control loop was probably not jammed during the accident sequence because there was no evidence of foreign material to cause such a jam and there were no nicks or gouges on the input linkage to indicate that a jamming material might have been present at impact. Further, the main rudder PCU’s external input linkage effectively covers (blocks) the opening between the input crank and the PCU body stop for the left rudder command direction, preventing jamming material from entering the area. The Safety Board considers that a dual slide jam is a less likely accident scenario than a jam of the secondary slide to the servo valve housing because the dual jam would require two extremely rare failures to exist in the servo valve at the same time. Analysis 274 Aircraft Accident Report of a secondary slide jam to the servo valve housing combined with overtravel of the primary slide. Other ADs issued by the FAA should result in improved operational procedures and pilot training programs for addressing the more general problem of uncommanded movement of the rudder, including rudder reversal. The Safety Board concludes that, when completed, the rudder design changes to the 737 should preclude the rudder reversal failure mode that most likely occurred in the USAir flight 427 and United flight 585 accidents and the Eastwind flight 517 incident. However, even with these changes, the 737 series airplanes (including the NG) remain susceptible to rudder system malfunctions that could be catastrophic. In its October 1997 briefings to the FAA, Boeing acknowledged that a rudder hardover on the 737-NG during the most critical phases of flight—takeoff and/or landing (which Boeing estimated as 60 to 90 seconds per flight)—would be catastrophic. Although this period of vulnerability appears limited, the takeoff and landing phases are when the pilot is most likely to use the rudder, particularly to apply a high-rate rudder input. Pilots can apply rudder inputs during the takeoff or landing ground roll as they use the rudder pedals for nosewheel steering; these inputs can occur at low altitude with a loss of engine power or during a turbulence encounter. Any malfunction resulting in uncommanded rudder motion during an engine failure or in turbulence at low altitude may be catastrophic because of the limited time, altitude, and roll control authority to regain control of the airplane.

[PROBABLE CAUSE] INTRA-COCKPIT COMMUNICATION AIR-GROUND COMMUNICATION TIME & TIME & SOURCE CONTENT SOURCE CONTENT 0137:41 CAM-3 ***. 0137:41 CAM-1a bravo for you, bravo for you, and what about the one that doesn't have? 0137:45 CAM-3 what do you mean by the one that doesn't have? 0137:46 CAM-1a the one that comes aboard and doesn't have any document that says that he should board. 0137:49 CAM-3 they're all reported to the station from Egypt, each with his own schedule. anyone who will come back should inform the station in Egypt. 0137:55 CAM-? [sound similar to clearing of throat] 0137:57 CAM-1a good, what about the one who comes back and nobody knows anything about him? 0138:01 CAM-3 I gave my instructions in the station... 0138:05 CAM-1a fine. that's what I told them, I, I, I'm not a trouble maker. I mean you know me. I don't need any headache. I'm telling them I am looking after all your interests. when you all are covered, neither the station nor any pilot can argue with you all, and should anything happen to you, you see, nobody could say I know nothing about you. INTRA-COCKPIT COMMUNICATION AIR-GROUND COMMUNICATION TIME & TIME & SOURCE CONTENT SOURCE CONTENT 0138:12 CAM [sound similar to cockpit door operating] 0138:28 CAM-3 why, why would I put anybody, Habashi? 0138:28 CAM-? yes. 0138:31 CAM-1a no, that's what I'm saying. 0138:32 CAM-3 even if somebody makes a mistake, we'll cover him too. 0138:35 CAM-1a I'm not saying anything.

[PROBABLE CAUSE] INTRA-COCKPIT COMMUNICATION AIR-GROUND COMMUNICATION TIME & TIME & SOURCE CONTENT SOURCE CONTENT 0136:05 HOT-2a no, not at all. 0136:08 CAM-1a Hisham is telling me that -- I don't know -- he has simulator tomorrow. 0136:11 HOT-2a execute... VNAV? 0136:18 CAM-1a you see, he is telling me he has simulator tomorrow... is that any good? he is coming straight from Los Angeles, you know, and he is going to have simulator. 0136:26 CAM [sound similar to cockpit door operating] 0136:38 [brief interruptions in audio on all CVR channels] 0136:42 CAM-1a what's your opinion about the crowding on this plane? 0136:44 CAM-3 what crowding? 0136:47 CAM-1a the crew that is aboard this plane. 0136:50 CAM-3 there is one going back that has simulator and another whom I have told in Egypt to come back on the same airplane from Los Angeles direct to Egypt for training... what difference does it make? INTRA-COCKPIT COMMUNICATION AIR-GROUND COMMUNICATION TIME & TIME & SOURCE CONTENT SOURCE CONTENT 0136:59 CAM-1a what? 0137:00 CAM-3 what difference does it make? 0137:00 CAM [sound of squeak] 0137:01 CAM-1a no, it doesn't make any difference to me. I, I am saying that for their sake. 0137:07 CAM-3 because those, you see, are not active. 0137:08 CAM-1a I know that. I didn't say anything. but I mean when you are surprised to see people about whom you don't know anything. It's good that you are here, you know. I don't know, I mean, anybody could... 0137:24 CAM-3 isn't the extra crew written in the general dec? 0137:28 CAM-1a I mean, as long as you are here, there is no problem, you are here. if you were not here, anyone could tell me, "Captain Hatem," should I say okay? 0137:38 CAM-3 no, I sent a telex concerning this operation. 0137:40 CAM-1a bravo for you.

[PROBABLE CAUSE] INTRA-COCKPIT COMMUNICATION AIR-GROUND COMMUNICATION TIME & TIME & SOURCE CONTENT SOURCE CONTENT 0143:16 CAM-2b I mean, you don't... don't... 0143:18 CAM-1a no. 0143:19 CAM-2b he's trying to please them. just so you know. 0143:22 CAM-1a oh yes, didn't he say if somebody made a mistake... 0143:24 CAM-2b yeah. 0143:25 CAM-1a you know, I don't harm anyone, I will cover for him. 0143:28 CAM-2b like I'm telling you. 0143:29 CAM-1a I told him it's all right. 0143:30 CAM-2b he's trying to please them. 0143:31 CAM-1a did you give the instructions that they should come aboard? he said to me, "I always send telex messages." that's it. 0143:37 CAM [sound of click] INTRA-COCKPIT COMMUNICATION AIR-GROUND COMMUNICATION TIME & TIME & SOURCE CONTENT SOURCE CONTENT 0143:38 CAM-2b he's trying to please them. as long as he pleases them, don't worry your head. 0143:47 CAM [sound of two thumps] 0143:50 CAM [sound of two clicks] 0143:52 CAM-1a since when do I worry my head... 0143:53 CAM [two consecutive whirring sounds similar to electric seat motor operating] 0143:54 CAM-1a I mean, you know. 0143:56 CAM-2b they, they... the word is that @ @ @ is making trouble. you are making trouble, that's what's being said by everyone. just so you know. 0144:06 CAM-1a I don't care. 0144:09 CAM-2b these guys are a bunch of #. and what's more, they are being controlled by the guy named, under the leadership of, of the guy named, the # named @, you know? 0144:22 CAM-1a ***. INTRA-COCKPIT COMMUNICATION AIR-GROUND COMMUNICATION TIME & TIME & SOURCE CONTENT SOURCE CONTENT 0144:24 CAM-2b don't worry your head. you're good hearted. 0144:26 CAM-1a I mean... you know me, it doesn't make any difference to me. but when kids start that way... 0144:35 CAM [sound of two clicks] 0144:37 CAM-1a I told him, I told him later that if I saw anyone aboard that didn't get permission from anybody and who also keeps going back and forth, even with permission, he won't fly with me.

[PROBABLE CAUSE] I won't fly with him, I'm not willing. I don't want to fly with anyone, even if it comes to me not flying, and if they can prevent me from flying, that's fine, what can I say to them? 0145:04 CAM-2b these kids are forming a clique with each other, just so you know, under the leadership of @ @. @ @ controls that group. he has @ ear, as well as... this kid is clever and cunning. 0145:10 CAM [brief interruptions in audio on all CVR channels] 0145:25 CAM-1a who? 0145:27 CAM-2b that's him. 0145:29 CAM [sound of click] INTRA-COCKPIT COMMUNICATION AIR-GROUND COMMUNICATION TIME & TIME & SOURCE CONTENT SOURCE CONTENT 0145:30 CAM-1a which one? 0145:35 CAM-2b that's enough. 0145:35 CAM-2b don't you have the flight report with you, Adel? 0145:37 CAM-2a I've got it, sir, I've got it. 0145:41 CAM-1a what's with you, why did you get all dressed in red like that? 0145:47 CAM [sound similar to cockpit door operating] 0145:48 CAM-1a when do you have simulator? 0145:49 CAM-5 Wednesday. 0145:50 CAM-1a what's today? 0145:51 CAM-5 Sunday... Saturday. 0145:56 CAM [sound of two clicks] INTRA-COCKPIT COMMUNICATION AIR-GROUND COMMUNICATION TIME & TIME & SOURCE CONTENT SOURCE CONTENT 0145:59 CAM-1a hey guy, why didn't you take tomorrow's plane? 0146:00 CAM-5 I tried (because the plane is the plane named thousand). 0146:03 CAM-1a why burn yourself out? 0146:05 CAM-5 because I'm a pilot (thousand). I made three (trips). I'm sick and taking medication. 0146:10 CAM-? maybe tomorrow? 0146:11 CAM-6 * do you all want something else? 0146:12 CAM-2b no, no, no. 0146:13 CAM-6 ***. 0146:15 CAM-2b that's really, really fantastic. 0146:17 CAM-6 ***. 0146:18 CAM-2b thanks a lot.

[PROBABLE CAUSE] INTRA-COCKPIT COMMUNICATION AIR-GROUND COMMUNICATION TIME & TIME & SOURCE CONTENT SOURCE CONTENT 0134:29 HOT-2a If you want to take these things, take them. 0134:31 CAM-4 (sure) 0134:33 CAM [sound similar to cockpit door operating] 0134:44 CAM-? *** . 0134:53 CAM [sound similar to cockpit door operating] 0135:08 HOT-2a what's new? what's wrong with you? 0135:13 CAM-1a that's nonsense. 0135:14 HOT-2a yeah, of course. 0135:14 CAM-1a quite frankly, that's nonsense. I don't accept it. 0135:17 [brief interruption in audio on all CVR channels] 0135:17 HOT-2a a market, it's a market in the cockpit. INTRA-COCKPIT COMMUNICATION AIR-GROUND COMMUNICATION TIME & TIME & SOURCE CONTENT SOURCE CONTENT 0135:20 CAM-1a I don't accept this situation. never... you can't call that flying. by Great God, call it anything else. that has nothing to do with flying, this business... I didn't want to cause trouble for one reason only: just because Hatem is here and he will see before him how the situation is. 0135:44 HOT-2a Captain, you... 0135:44 HOT-2a all of this because of this good-for-nothing guy named Ayyad... 0135:46 CAM-1a what? 0135:47 HOT-2a this Ayyad who keeps saying Zulu Alpha... Zulu Alpha. 0135:49 CAM-1a I am aware of it. 0135:50 HOT-2a he sat here and entered three nine. 0135:52 CTR2 EgyptAir nine ninety climb and maintain flight level three three zero, cleared direct DOVEY. 0135:57 RDO-2a three three zero, direct DOVEY, EgyptAir nine nine zero. 0136:02 CAM-1a between me and you this is not work, by God seriously.

[(b) PROBABLE CAUSE] C, 20591 On September 8, 1974, Trans World Airlines Flight 841, a Be707#331B, crashed in the Iontan Seca, about 50 miles west of Cephatonia, Greeces The aircraft was on a flight from Tel Aviv, Israel, to New York, New York, with scheduled stops in Athens, Greece, and Rone, Italy. The results of the laboratory examination of certain items in the recovered flotsam establish conclusively that the detonatton of a high order explosive took place in the afreraft's aft cargo compartment, Subsequent to the accident 4 was éetermined that an aft cargo conpartment fire on a similar flight on August 26, 1974, was caused by a malfunctioning explosive device contained in a suitcase, In both instances, the passengers! checked bagzape at the Last board= ing point was not examincd, nor was this required, Trans Horld Airlines’ procedures now include the examination of checked baggape at that boarding point, The National Transportatton Safety Borrd is aware of the problems in maintaining, on adequate level of aircraft security with» out undue costs, delayr, or passenger irritation, especially when an afr carrier operates in other countries, Since aircraft security, in tost cases, is a joint responsibility of the air carrier, the aire port authority, ard the regulatory agency involved, it ta apparent that close coordination anony all parties involved tg a prerequisite for the effectivencas of the sccurity program, The Federal Aviation Administration's Aviation Security Technical Asatatance Program playa avital role fn adapting security programs to the needs of time and locale, Although many nations have already availed thenselvas of this Program, it has not yet reached all countrics where Amrtean flap carriers make echeduled atopas 1457 APPENDIX H Honorable Alexander P, Butterfield (2) The Safety Roard notes that, with the exception of the FAn's Regional headquarters in Brussels, Belgium, for Europe, Africa and Middle East, all regional headquarters have security offices, The establishment of such an office in your Brussels headquarters would provide a mucheneeded focal point for the coordination of aircraft security measures in the area served by that headquarters, Although aircraft sabotage can take many forms, it appears that, in most cases, soma type of high explosive is involved, The Safety Board is aware of the ongoing research in the deve lopment of explosives detection equipment and believes that the use of suitable detection equipment would not only simplify examination procedures but serve as a deterrent, Accordingly, the National Transportation Safety Board recommends that the Federal Aviation Administration?! 1, Recemphasize to the nations served by American flag carriers the importance of participating in the Aviation Security Technical Agsistance Program, Establish an Aviation Security Office in the Federal Aviation Administration's Europe, Africa and Middle Bast Regional Headquarters in Brussels, Belgium, Expedite the development and use of suitable explosives detection equipment to preclude the introduction of explosive devices on board an aircraft, Ensure that the afreraft security programs of UeS. air carriers, as prescribed by 14 CFR 121,538, contain provisions that are more responsive to high risk situations in international as well as domestic operations, Our technical staff is avatlable for any further they may be able to provide, /K By’ John Hy, Reed Chaitman

[CONCLUSIONS > FINDINGS] The fasten seatbelt signs and public address equipment were in working order. The occupants were injured because some were not securely restrainec! in their seats and because some were hit by loose articles in the cabin. The flight attendants and passengers had sufficient time to secuve themselves in their seats before the severe turbulence encounter. The means for insuring restraint of the Passenger service carts at the serving stations in passenger aisles needs improvement. Neither the flighterew nor the controller was aware of the convective SIGMET's that had been issued after Flight 965 left Trinidad. The controller provided adequate information and instructions to other flights In the area once he became aware of the PIREP on turbulence. The manner in which the FAA distributed information regarding implementation of the HIWAS program was inadequate. The current HIWAS program is not adequate because the FAA did not consider maximum reception altitudes, the location of traveled preferential jet routes, and trans-Atlantic and trans~Pacific traffic. 19. The ability to use sophisticated on-board navigational computers successfully with the HIWAS program needs to be established. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of the accident was ar encounter with severe clear air turbulence produced by the intrusion of thunderstorm cells into strong winds aloft. 4. RECOMMENDATIONS As a result of the accident, the National Transportation Safety Board recommended that the National Oceanic and Atmospheric Administration: Advise its weather forecasters to be alert for situations where there is a jet stream or strong upper level winds in association with lines of developing or developed thunderstorms which may produce an area of ° a 4 + 3 5 “4 ¢ 4 x i C.D ee a ~25- severe clear air turbulence, and to tasue appropriate warnings of this potential turbulence to pilots through area forecasts, SIGMETs or other poroprlate means of communication. (Class I, Priority Action) -~that the Federal Aviation Administration: Postpone nationwide implementation of the Hazardous inflight Weather Advisory Service Progearam at Air Traffie Control Centers untt] the broadcasting procedures are improved and program information is disseminated widely. (Class I, Priority Action) (A-84-111) Designate communication frequencies within the 118-135 MHz band for each Air Route Traffie Control Center to broadcast Hazardous inflight (eater Advisory Service information. (Class 1, Priority Action) Develop procedures similar to those currently used in terminal areas for Automatic Terminal Information Service, for flightcrews to monitor an individual facility's Hazardous Inflight Weather Advisory Service frequency and to inform the controller/facility on initial contact that fhe flight has the current HIWAS information. (Class N, Priority Action) A~84-112 During a transition Period following the implementation of Hazardous Inflight Weather Advisory Service, require Air Traffic Controllers to advise flighterews when critical safety information is being made available through HIWAS, For example, ARTCC controllers should be required to advise flights upon initial contact "significant weather information available on HIWAS." (Class Ny Priority Action) (A-84-114) Institute e@ program to ensure that changes to ATC operations and communications procedures, meang to disseminate aviation weather information, ete., are publishea in a manner to directly reach all users of the National Airspace System. (Class Il, Priority Action) (A-84-113) Also as a@ rasult of its investigation, the National Transportation Safety Board suggested that the Canadian Aviation Safety Board recommend to the Canadian Air Transportation Administration that it: Require Air Canada fo initiate a daily inspection program to assure that each passenger servive cart (PSC) locking mechanism is undamaged and ean be properly aligned with the floor-mounted anchor pin until a positive lock indicator is Installed or a more reliable means of positioning and anchoring the PSC is designed and installed.

[FINDINGS] Pilgrim Airlines training involved a different procedure for flightcrews to follow when a propeller autofeathers during takeoff than did the airline's flight manual. Neither procedure was an optional one. The flight attendant manual had conflicting information with regard to the use of seatbelts for infants and children, and was unclear with regard to the duties and responsibilities for blocking of seats and weight restrictions for vargo/baggage. The flight attendant manual contained no instructions to the flight attendant to remain seated with their restraints fastened until the airplane's movement had stopped and no instructions governing the proper stowage of oversize galley service items. When some passengers attempted to assume one of the two brace positions depicted on the safety card, i.e., arms braced on the seatback in front and head resting on arms, the seatbacks folded over, and passengers had little time in which to assume alternate or brace positions before the airplane struck the runway. Passengers! Injuries resulted predominately from inertia forces on inittal impact, which caused cervical-sacral strains end minor contusions when passengers struck adjacent seats. The captain should not have been allowed to reenter the airplane alone to shutoff the electrical power, but should have been accompanied by a properly equipped firefighter with a charged handline. Federal Aviation Administration surveillance of Pilgrim Airlines did not provide adequate monitoring of its operational programs. 3.2 Probable Cause “ The National Transportation Safety Board deiermines that the probable cause of this accident was the flighterew failure to use engine anti-ice on the inbound flight to JFK, the captain's failure to conduct a thorough prerlignt inspection, and the flightcrew's decision to use engine anti-fce on takeoff from JFK which lead to power lossos on both engines, 4. RECOMMENDATIONS As a result of its investigation, the National Transportation Safety Board recommends that the Federal Aviation Administration: Issue instructions to air currier Principal Maintenance Inspectors responsible for F-27 airplanes to examine underwing emergency exits for interference from adjacent passenger seats, and where interference is found, tc direct air carriers to eliminate the interference within a specified ime. (Class I, Priority Action) (A-84-128) Issue instructions to air carrier Principal Maintenance Inspectors responsible for F~27 airplanes to require air carriers to install, within a specified time, an FAA-approved means to prevent the hinge pins from coming free of their hinges on the door between the forward cabin and the cargo compartment or to remove that door. (Class Il, Prior'ty Action) (A-84-129) Issue instructions to air carrier Principal Operations Inspectors to review the passenger safety information cards of their respective carriers to assure that any depicted bracing position, utilizing the seatback for support, in fact can be used; and to require deletion of this bracing position from the safety infcrmation cards on those airplanes that are equipped with seats that have foldover seatbacks. (Class II, Priority Action) (A-84-130) I Issue instructions to the alr carrier Principal Operations Inspector to require revision of the flight attendant manuals of Filgrim Airlines to incorporate elear, concise, and unambiguous operating instructions, and to conform to accepted industry standards, and to require that the training program for crewmembers be consistent with the manuals. (Class Il, Priority Action) (A-84-131) Issue ine‘tcuctions to air carrier Principal Operations inspectors to require that flight attendent training programs and manuals of air carriers address adequately the need to stow galley service items in approved compartments and to include, during thelr in-servics inspections, increased surveillance of the proper pre-flight and preers stowage of galley service itxms. (Class I, Priority Action) A- ~1li Pe BTR GOR NESTRAE A PI?

[ANALYSIS] Such an attitude appears to have applied also to his violating velatively routine company procedures. For example, the captain created an incident as a nonreveriue passenger when his baggage did not arrive at the airport In an instance in which he had claimed ‘hat his companion on the flight was his wife when she was not. He twice violatud company operating procedures by fueling an airplane himself and foading a passenger with an engine operating; both instances also supporting his reputation as a pilot who liked to hurry. 2.4 Cocaine The literature on cocaine indicates that its use is still evelving in this country, both in the type of uta, habitual vs. occasional, as weil as the quality or purity of the drug. Cortainty, public perception of the use of the drug has changed over the last few years with the cocaine-related Rie RRA Hitt ale SUE ARIAS) CLO RN ARM ON ae Le DI RED BINNIE CO SR ei a SE Be cosine Fst SNA at tes, Bi FAR GS a CEN erie ae inoaeal 30 deaths and injuries of public figures. However, as this accident demonstrates, its use by pilots poses a threat to the safety of the flying public. To exacerbate the problem, cocaine use is difficult to detect, even by individuals who interact daily with an abuser. Moreover, the behavioral manifestations of cocaine use, which are often quite subtle, are affected by several tactors in addition to dosage. These include the method of ingestion, tolerance to the drug, and other factors which interact to create the variability in behavioral and physiological effects following both cocaine use and withdrawal from its use. Further, the complexity of the effects of cocaine ingestion and subsequent perforvmance impairment exteid to a host of licit and illicit drugs. As a result, this accident demonstrates both the danger of cocaine use in aviation and the difficulty faced by the aviation community in attempting to contro! that use. The Safety Board previously examined the use of illicit drugs in its investigation of an airplane accident at Newark, New Jersey on March 30, 1984.!2 As a result of that accident, the Safety Board recommended that the FAA: A-84-95 In coordination with the Cffice of the Secretary, U.S. Department of Transportation, institute appropriate research to further the understanding of potential effects on pilot performance of both licit and illicit drugs, in both therapeutic and abnormal levels, and actively disseminate those fi ndings. The FAA responded that a working group with the Department of Transportation (DOT) was created and a literature search was funded and began. On December 29, 1988, the FAA informed the Safety Board that the literature seaych had been completed and that distribution of the report, Date Available on the impact of Drug Use on Transportation Safety, would be accomplished through the regional flight surgeons.

[PROBABLE CAUSE] Contributing to the injuries was flight attendant A’s incorrect assessment of his time of useful consciousness, which led to his failure to follow procedures requiring immediate donning of an oxygen mask when cabin pressure is lost. BY THE NATIONAL TRANSPORTATION SAFETY BOARD DEBORAH A.P. HERSMAN ROBERT L. SUMWALT Acting Chairman Member CHRISTOPHER A. HART Member MARK R. ROSEKIND Member EARL F. WEENER Member

[ANALYSIS] The passenger seats were equipped for underseat stowage of carryon baggage in accordance with 14 CFR 135.87, but one passenger said he had his attache case on his lap during takeoff. Although 14 CFR 135.87(¢)(6) requires carryon baggage to be stowed before takeoff and before landing, there is no requirement that flighterews inform passengers of the requirement or that safety cards contain this information. Neither the pretakeoff briefing nor the safety briefing cards on the accident aircraft mentioned requirements for carryon baggage. This type of safety information becomes most important in commuter operations because operators usually are not required to have onboard a cabin attendant who would insure that the passengers have their seatbelts fastened, that there is no smoking, and that carryon baggage is stowed beneath seats. Consequently, the Safety Board believes that the pretakeoff briefing and safety briefing cards should be amended to inform commuter passengers of the carryon baggage stowage requirement. Title 14 CFR 135.117 requires that the pretakeoff briefing of passengers inelude information concerning the use of seatbelts. The Pilgrim Airlines flighterew pretakeoff briefing did not address seatbelt use. Likewise, the seatback safety briefing ecards did not address seatbelt usage or operation of the seatbelts as supplemental information. The Safety Board believes that the oral briefing should comply with the regulatory requirement and that the seatback safety briefing cards should provide supplemental information on the operation and the use of the seatbelts. The safety briefing card also did not show the location of the overhead escape hatch located in the rear cabin. The Safety Board believes that more comprehensive surveillance by FAA inspectors would have discovered the obvious discrepancies in the safety card and the lack of instructions for the use and operation of seatbelts in the pretakeoff briefing. The FAA's surveillance of air taxi and commercial operators which operate under 14 CFR Part 135 should certainly place heavy emphasis on occupant safety and safety equipment. The rapid growth of the commuter industry, as discussed in the Safety Board's special study on commuter airline safety, 5/ clearly requires that FAA inspectors be aware of the need for operators to conform to all applicable requirements in the Federal Aviation Regulations under which they operate. 2.4 Survival ets Fire injuries were sustained by the captain, first officer, and two passengers. The crew sustained their injuries in flight and before escaping from the aircraft after it landed. The captain's injuries were more severe and extensive because the fire appeared to have been more extensive in the left cockpit and left forward cabin. The first officer was wearing several layers of clothing and was thus afforded more protection from radiant heat and direct flames. None of the surviving passengers was incapacitated during the in-flight fire. The rapid evacuation of the passengers was possible because the cabin remained intact and no debris hampered access to the forward right emergency exit or to the opening in the rear of the cabin.

[ANALYSIS] The Safety Board’s aircraft performance study of the tire marks on runway 4L from the accident airplane (see section 1.18.7) indicated that it departed the left edge of the runway with a shallowing leftward veer. This evidence implies that the captain was beginning to regain control of the airplane when it left the runway. The simulation study results indicated that tiller inputs alone would have been incapable of this recovery of control. When interviewed after the accident, the captain recalled that he had applied increasing amounts of right rudder as the airplane veered to the left. However, based on the consistent effectiveness of rudder inputs in the simulation study and the tire mark evidence that directional control was being regained at the runway’s edge, the Safety Board concludes that the captain of flight 41 first relied on right tiller inputs as the airplane continued to veer left, then applied insufficient or untimely right rudder inputs to effect a recovery. 2.2.3 Timeliness of the Rejected Takeoff In his postaccident interview with the Safety Board, the captain stated that after noting the airplane’s failure to respond to his initial input of right rudder, and before deciding to reject the takeoff, he applied additional right rudder and tiller steering inputs. He then described 43 his attempts to reject the takeoff by retarding power to idle and applying maximum braking, right rudder, and nosewheel steering input. Thus, instead of rejecting the takeoff immediately after experiencing difficulty obtaining directional control, the captain continued to attempt to regain directional control with progressively greater rudder and nosewheel steering inputs. Because the FDR was not working, the Safety Board did not have sufficient information to measure the delay between the first indication of loss of control and the captain’s subsequent reduction of engine power. However, some measure of the extent of the delay can be gained from the simulation and performance studies. The simulation study showed that loss of directional control began at the relatively slow airspeeds when the aerodynamic rudder had not yet become effective (less than 50 knots), while the aircraft performance study showed that the accident airplane departed the left side of the runway at a relatively high speed (approximately 97 knots). The captain stated that he reduced power while the airplane was still on the runway, and that he had no recollection of subsequently reapplying power. However, the Safety Board’s CVR spectrum analysis clearly indicated that the thrust was partially reduced and then reapplied in significant amounts as the airplane left the runway. Physical evidence from the engines and flightcrew statements confirmed that the engine rpm increase recorded on the CVR was not an engagement of reverse thrust. Because the CVR ceased recording shortly after the reapplication of power to the engines, the Safety Board was unable to determine the amount of time that the airplane traveled off the runway under significant power. However, based on the spectrum analysis of engine sounds on the CVR, the Safety Board determined that the captain abandoned his attempt to reject the takeoff, at least temporarily, by restoring forward thrust.

[ANALYSIS] Given that it is very unlikely that the captain did not try to control the airplane’s tendency to weathervane into the crosswind, and given the consistent controllability of the airplane under accident conditions when the tiller was not used (during the simulation study), the Safety Board concludes that the captain’s failure to correct the airplane’s deviation from the centerline resulted from his overcontrolling the nosewheel steering through the tiller. This conclusion is supported by the captain’s statement that he added increasing amounts of tiller steering input during the loss of control sequence and departed the runway still holding full right tiller. The Safety Board was unable to determine with certainty the event that precipitated the captain’s overcontrol with tiller inputs. Simulation study results suggest that the B-747 has a tendency to react to crosswinds at very slow airspeeds with an initial, slight downwind drift. It would have been natural for the captain to have reacted to this slight 42 deviation with a tiller input, because the deviation would have occurred at a slow airspeed as the airplane was just beginning its takeoff roll. However, there could have been a number of additional reasons for why the captain applied steering inputs through the rudder or tiller at the start of the takeoff roll. These include a line-up that was slightly off the runway centerline, a wind gust, or a slight thrust imbalance from one or more engines as they accelerated to takeoff power. Still, regardless of the reason for beginning the control inputs, the simulation study indicated that the runway deviation was unlikely to have precipitated a loss of control without excessive steering inputs through the tiller. It is logical that overcontrol of the tiller on any aircraft would be more likely on a slippery runway than a dry runway, because airplane heading is less responsive to tiller inputs in slippery conditions. When a pilot makes a tiller input and does not obtain the expected reaction from the airplane, it is possible that the pilot will, at least initially, provide additional input to obtain the expected reaction. The lag in airplane response followed by additional control input could result in overcontrol of the tiller to the extent that the nosewheel exceeds its critical angle and loses traction. The simulation study also demonstrated that at least enough rudder effectiveness was obtained by 50-80 knots airspeed to shallow the simulator’s leftward veer before it departed the runway. In most simulations, directional control could be regained by timely use of the rudder. Given the effectiveness of rudder inputs in controlling heading deviations in the simulation study, the Safety Board sought to understand why the captain of the accident airplane was unable to recover directional control before the airplane departed the left side of the runway. The Safety Board’s aircraft performance study of the tire marks on runway 4L from the accident airplane (see section 1.18.7) indicated that it departed the left edge of the runway with a shallowing leftward veer. This evidence implies that the captain was beginning to regain control of the airplane when it left the runway. The simulation study results indicated that tiller inputs alone would have been incapable of this recovery of control.

[ANALYSIS] Although 5 minutes before the accident ATC changed the departure runway to 31L for traffic following flight 41, the Safety Board recognizes that the captain’s decision to use runway 4L was based on the limited information available to him at the time. Further, air traffic controllers were not required to offer flight 41 the option of switching to runway 31L, once the airplane was established holding short at runway 4L. Based on the absence of definitive runway friction measurements for runway 4L, reported winds of less than 15 knots (the maximum recommended crosswind component for B-747 takeoffs on slippery runways), the flightcrew’s reports of acceptable visibility down the runway, and the reported unavailability of the alternative runway 41 31L, the Safety Board concludes that the captain’s decision to attempt the takeoff on runway 4L was appropriate. 2.2.2 The Attempted Takeoff and Loss of Control Flight 41 attempted its takeoff under crosswind conditions with a runway contaminated with packed snow and patchy ice. At the approximate time of the takeoff attempt, there were crosswinds of 10-12 knots. Gusts of up to 22 knots were reported in the general area near the time of the accident. Asymmetric thrust (for example, inadequate thrust from the No. 1 engine) could have resulted in the loss of directional control experienced by flight 41. In the absence of a cross-check of other engine instruments, a malfunctioning EPR indicator could have led the flightcrew to unknowingly set inadequate thrust for the No. 1 engine. However, given the flight engineer’s recollections of evenly matched engine acceleration and consistent EPR and N1 indications from all four engines, and the absence from the CVR of flightcrew discussions of abnormal throttle alignment, the Safety Board concludes that asymmetric thrust was not a factor in the loss of directional control. Having verified the realism of the Boeing engineering flight simulator in reproducing the ground handling characteristics of the B-747 on slippery runways, the Safety Board applied the findings of its August 8, 1996, flight simulation study to the circumstances and events in this accident. In all simulations in which the pilot did not use the nosewheel steering tiller for directional control (including those conducted with icy runway conditions and winds gusting up to 40 knots), the simulated airplane was controllable along the runway centerline. In contrast, when pilots attempted to maintain the runway centerline using the tiller under slippery runway conditions with a 12-knot crosswind, a slight overcontrol at the very beginning of the takeoff roll repeatedly led to the loss of traction and steering capability from the nosewheel, followed by the loss of directional control. Given that it is very unlikely that the captain did not try to control the airplane’s tendency to weathervane into the crosswind, and given the consistent controllability of the airplane under accident conditions when the tiller was not used (during the simulation study), the Safety Board concludes that the captain’s failure to correct the airplane’s deviation from the centerline resulted from his overcontrolling the nosewheel steering through the tiller.

[ANALYSIS] Because the CVR ceased recording shortly after the reapplication of power to the engines, the Safety Board was unable to determine the amount of time that the airplane traveled off the runway under significant power. However, based on the spectrum analysis of engine sounds on the CVR, the Safety Board determined that the captain abandoned his attempt to reject the takeoff, at least temporarily, by restoring forward thrust. The Board’s aircraft performance study indicated that as a result of the reapplication of thrust, the airplane continued to accelerate as it approached the edge of the runway. 2.2.4 B-747 Slippery Runway Operating Procedures Because the Safety Board recognized that on a slippery runway, directional control of the B-747 could be lost rapidly by overcontrol of the tiller, it evaluated the existing procedures established by Tower Air and Boeing for operating the B-747 on slippery runways. As a result of the Tower Air procedure to guard the tiller during takeoff until 80 knots, the captain was ready to use the tiller during the beginning of the takeoff roll. Tower Air and Boeing procedures urge pilots to use the rudder and rudder pedal steering during takeoff. However, B-747 procedural information produced by both the airline and the manufacturer permit the tiller to be used at the beginning of the takeoff. In its 1994 Standards Memo, Tower Air stated, “Use of the tiller is not recommended unless rudder pedal steering is not sufficient during the early takeoff roll.” Boeing stated in its Flight Crew Training Manual for the B-747, “Do not use nosewheel tiller during takeoff roll unless required initially due to crosswind.” The Safety Board is concerned that these procedures encourage use of the 44 tiller at the beginning of the takeoff roll, during which the Safety Board’s simulation study found the B-747 to be most susceptible to loss of control on slippery runways. The Safety Board concludes that current B-747 operating procedures provide inadequate guidance to flightcrews regarding the potential for loss of directional control at low speeds on slippery runways with the use of the tiller. The Safety Board believes that the FAA should require modification of applicable operating procedures published by Boeing and air carrier operators of the B-747 to further caution flightcrews against use of the tiller during slippery runway operations, including low-speed operations (for airplanes equipped with rudder pedal steering) and to provide appropriate limitations on tiller use during these operations (for airplanes not equipped with rudder pedal steering). The Safety Board was informed by Tower Air after the accident that it had reevaluated and eliminated its standard procedure of guarding the tiller during the takeoff roll through 80 knots. The Safety Board concludes that this procedural change by Tower Air will make overcontrol of the tiller less likely for its own operations; however, other air carrier operators of the B-747 may need to make similar changes to their procedures.

[ANALYSIS] The Board is concerned that air carrier B-747 pilots currently are not able to obtain needed training on slippery runway procedures, including proper tiller and rudder techniques, because training simulators have not incorporated the latest ground handling model (such as that implemented on the Boeing engineering simulator). Further, although existing flight test data on slippery runway handling characteristics are limited, the increasing use of high capacity FDRs and quick access maintenance recorders enables data on slippery runway handling to be obtained from actual line flying experience. Many B-747-400 models are equipped with these recorders. The Safety Board concludes that improvements in the slippery runway handling fidelity of flight simulators used for B-747 pilot training are both needed and feasible. Consequently, the Safety Board believes that the FAA should evaluate B-747 simulator ground handling models and obtain additional ground handling data, as required, to ensure that B-747 flight simulators used for air carrier flightcrew training accurately simulate the slippery runway handling characteristics of the airplane. The Safety Board also believes that after completing this evaluation, the FAA should issue an FSIB urging POIs assigned to air carrier operators of the B747 to enhance simulator training for slippery runway operations, including limitations on tiller use and instructions for rudder use during the takeoff roll. 2.2.6 Summary of Flightcrew Actions and Decisions The captain’s use of the tiller control for nosewheel steering during the takeoff roll, combined with his untimely or inadequate use of rudder inputs, allowed the loss of directional control to develop. As this occurred, the airplane’s deviation from the centerline and its unresponsiveness to steering inputs provided cues that, regardless of the adequacy of existing procedures and training methods, should have prompted the captain to reject the takeoff more quickly than he did. Therefore, the Safety Board concludes that the captain’s failure to reject the takeoff in a timely manner was causal to the accident. Still, better procedures for operating the B-747 under slippery runway conditions and improved ground handling fidelity of the flight simulators used for B-747 pilot training could have better prepared the captain for handling the situation that confronted the accident flight.

[ANALYSIS] Further, there are no means to compare measurement standards or translate the data into aircraft performance. A key issue is that no significant progress has been made in correlating stopping distance data from airplane manufacturers’ flight tests and calculations with the friction values obtained from measuring devices. An outcome of these correlations could be the establishment of objective standards for air carrier operations on slippery runways, perhaps extending to the establishment of appropriate minimum runway friction levels for operational use. The Safety Board concludes that the circumstances of this accident indicate that the issue of correlating airplane stopping performance with runway friction measurements should be revisited by the Government and the air transportation industry. Consequently, the Safety Board believes that the FAA should require the appropriate Aviation Rulemaking and Advisory Committee to establish runway friction measurements that are operationally meaningful to pilots and air carriers for their slippery runway operations (including a table correlating friction values measured by various types of industry equipment), and minimum coefficient of friction levels for specific airplane types below which airplane operations will be suspended. 55 3. CONCLUSIONS 3.1 Findings 1. The flightcrew was properly certificated and qualified in accordance with applicable regulations and company requirements. 2. The air traffic control personnel involved with the flight were all properly certificated and qualified. 3. The airplane was properly certificated, equipped, and maintained (with the exception of the flight data recorder system) in accordance with approved regulations. The weight and balance were within allowable limits. 4. The captain’s decision to attempt the takeoff on runway 4L was appropriate. 5. Asymmetric thrust was not a factor in the loss of directional control. 6. The captain’s failure to correct the airplane’s deviation from the centerline resulted from his overcontrolling the nosewheel steering through the tiller. 7. The captain of flight 41 first relied on right tiller inputs as the airplane continued to veer left, then applied insufficient or untimely right rudder inputs to effect a recovery. 8. Current Boeing 747 operating procedures provide inadequate guidance to flightcrews regarding the potential for loss of directional control at low speeds on slippery runways with the use of the tiller. 9. The procedural change by Tower Air to reevaluate and eliminate its standard procedure of guarding the tiller during the takeoff roll through 80 knots will make overcontrol of the tiller less likely for its own operations; however, other air carrier operators of the Boeing 747 may need to make similar changes to their procedures. 10. Current Boeing 747 flight manual guidance is inadequate about when a pilot should reject a takeoff following some indication of a lack of directional control response. 11. Improvements in the slippery runway handling fidelity of flight simulators used for Boeing 747 pilot training are both needed and feasible. 12. The captain’s failure to reject the takeoff in a timely manner was causal to the accident. 13. The inadequate Boeing 747 slippery runway operating procedures developed by Tower Air and the Boeing Commercial Airplane Group, and the inadequate fidelity of B-747 56 flight training simulators for slippery runway operations, contributed to the cause of this accident. 14.

[ANALYSIS] The Safety Board concludes that this procedural change by Tower Air will make overcontrol of the tiller less likely for its own operations; however, other air carrier operators of the B-747 may need to make similar changes to their procedures. Consequently, the Safety Board believes that the FAA should issue a flight standards information bulletin (FSIB) to POIs assigned to air carriers operating the B-747, informing them of the circumstances of this accident and requesting a review and modification, as required, of each air carrier’s takeoff procedure regarding pilot hand position with respect to the tiller. The Safety Board recognizes that it may be a natural reaction for a pilot to persevere in a takeoff attempt when faced with an apparently minor hesitation of an airplane to respond to rudder input. However, the circumstances of this accident indicate that during takeoff in a B-747 on a slippery runway, the pilot must abort at the very first indication of a directional control loss. The Boeing B-747 Operations Manual and Tower Air B-747 Flight Manual direct pilots who are performing takeoffs on slippery runways to immediately reject the takeoff if deviations from the runway centerline cannot be controlled. While this accident demonstrates the soundness of this advice, the accident also indicates that the provisions in these manuals are not adequately specific, particularly in their references to deviations that “cannot be controlled.” Tower Air’s chief of flight standards suggested a criterion for rejecting takeoffs under slippery runway/crosswind conditions that may be useful for pilot decisionmaking in the future. He linked the takeoff rejection decision to the recommended procedure of limiting rudder pedal steering input to one-half full travel to get optimal cornering friction. He indicated it was clear that if a pilot could not control the airplane with one-half rudder pedal travel, the takeoff should be rejected. This advice may be operationally useful for all B-747 pilots, if it can be verified by the FAA and aircraft manufacturer. The Safety Board concludes that current B-747 flight manual guidance is inadequate about when a pilot should reject a takeoff following some indication of a lack of directional control response. Consequently, the Safety Board believes that 45 the FAA should require Boeing to develop operationally useful criteria for making a rapid and accurate decision to reject a takeoff under slippery runway conditions; then require that B-747 aircraft flight manuals, operating manuals, and training manuals be revised accordingly. 2.2.5 Training Simulators for B-747 Slippery Runway Operations The air carrier and FAA pilots who participated in the August 8, 1996, simulation study believed that the Boeing engineering simulator had more realistic ground handling performance than the simulators Tower had provided for pilot training.

[PROBABLE CAUSE] Inadequate Boeing 747 slippery runway operating procedures developed by Tower Air, Inc., and the Boeing Commercial Airplane Group and the inadequate fidelity of B-747 flight training simulators for slippery runway operations contributed to the cause of this accident. The captain’s reapplication of forward thrust before the airplane departed the left side of the runway contributed to the severity of the runway excursion and damage to the airplane. 58 4. RECOMMENDATIONS As a result of the investigation of this accident, the National Transportation Safety Board makes the following recommendations: --to the Federal Aviation Administration: Require modification of applicable operating procedures published by the Boeing Commercial Airplane Group and air carrier operators of the B-747 to further caution flightcrews against use of the tiller during slippery runway operations, including low-speed operations (for airplanes equipped with rudder pedal steering) and to provide appropriate limitations on tiller use during these operations (for airplanes not equipped with rudder pedal steering). (A-96-150) Issue a flight standards information bulletin to principal operations inspectors assigned to air carriers operating the B-747, informing them of the circumstances of this accident and requesting a review and modification, as required, of each air carrier’s takeoff procedure regarding pilot hand position with respect to the tiller. (A-96-151) Require the Boeing Commercial Airplane Group to develop operationally useful criteria for making a rapid and accurate decision to reject a takeoff under slippery runway conditions; then require that B-747 aircraft flight manuals, operating manuals, and training manuals be revised accordingly. (A-96-152) Evaluate Boeing 747 simulator ground handling models and obtain additional ground handling data, as required, to ensure that B-747 flight simulators used for air carrier flightcrew training accurately simulate the slippery runway handling characteristics of the airplane. (A-96-153) After completing this evaluation, issue a flight standards information bulletin urging principal operations inspectors assigned to air carrier operators of the Boeing 747 to enhance simulator training for slippery runway operations, including limitations on tiller use and instructions for rudder use during the takeoff roll. (A-96-154) Develop certification standards for the installation of secondary galley latches; then use those standards to conduct an engineering review of secondary galley latches on all transport-category 59 aircraft.

[FINDINGS] Flightcrew comments on the CVR prior to the accident suggested that they were operating under self-induced pressure to make a landing curfew at the destination airport, and that this may have influenced their decisionmaking. 11. The flight engineer improperly determined the Vmcg speed, resulting in a value that was 9 knots too low. Neither the captain nor the first officer detected the error. 12. During the first attempted takeoff, the captain was not able to maintain directional control because he applied high power to the asymmetrical engine too soon, and he rejected the takeoff. During the taxi back for a second takeoff, he and his crewmates did not properly analyze the reasons for the loss of control. 13. The captain agreed to modify the three-engine takeoff procedure by - allowing the flight engineer to advance the throttle on the asymmetrical engine, a deviation of the prescribed procedure.. The captain was unable to maintain directional control on the second takeoff, decided not to reject the takeoff, and rotated the airplane early i in an attempt to take off prior to departing the paved runway surface. 14. FAA oversight of ATI was inadequate because the ATI POI and the geographic inspectors were unable to effectively monitor domestic crew training and international operations, respectively. 15, Existing FAR Part 121 flight time limits and rest requirements that pertained to the flights that the flightcrew flew prior to the ferry flights did not apply to the ferry flights flown under FAR Part 91. This permitted a substantially reduced flightcrew rest period when conducting the nonrevenue ferry flights. . 78 16. Current one-engine inoperative takeoff procedures do not provide adequate rudder availability for correcting directional deviations during the takeoff roll compatible with the achievement of maximum asymmetric thrust at an appropriate speed greater than ground minimum control speed. 79 3.2 Probable Cause The National Transportation Safety Board determines that the probable causes of this accident were: ) (1) the loss of directional control by the pilot in command during the takeoff roll, and his decision to continue the takeoff and initiate a rotation below the computed rotation airspeed, resulting in a premature liftoff, further loss of control and collision with the terrain. (2) the flightcrew's lack of understanding of the three-engine takeoff procedures, and their decision to modify those procedures. (3) the failure of the company to ensure that the flightcrew had adequate experience, training, and rest to conduct the nonroutine flight. Contributing to the accident was the inadequacy of FAA oversight of ATI and FAA flight and duty time regulations that permitted a substantially reduced flightcrew rest period when conducting a nonrevenue ferry flight under 14 CFR Part 91. 80

[ANALYSIS] This flame was the result of an engine compressor surge caused by disrupted airflow into the engine during the high angle of attack flight of the airplane immediately after liftoff. The airplane was inspected and maintained according to currently accepted practices, and all airplane systems appeared to be operating normally during the accident sequence of events. Available engine power was sufficient to successfully complete the takeoff, had the correct procedures been used by the flightcrew. The presence of the tire marks on the runway indicates that the thrust asymmetry of the three-engine takeoff exceeded the capability of the rudder (and the nose wheel steering, if used) to maintain directional control. It is not known whether the captain utilized the steering tiller during any portion of the takeoff attempts. In addition, data available from Douglas show that the engine power of the No. 4 engine, as indicated on the CVR, would have exceeded the capability of full rudder and nose wheel steering to maintain directional control. On both takeoff attempts, tire marks began early in the takeoff roll. This is consistent with data from the CVR showing that the thrust on the No. 4 engine was increased too quickly after brake release, resulting in excessive thrust asymmetry during the accident takeoff. FDR heading data and the presence of nose tire marks almost 10 feet to the right of runway centerline on the second takeoff attempt suggest that the captain may have steered the airplane to the right to provide the airplane more room to maneuver as the thrust from the No. 4 engine was increased, anticipating possible problems maintaining directional control. . 2.2 Airplane Systems 2.2.1 Brakes, Landing Gear and Tires The brake stacks were compressible and showed no evidence of melting, fusing or exposure to fire. In addition, there was no evidence of damage or malfunction to the nose wheel steering system, tires, or anti-skid system. No flat spots were seen on the tires, and no melted fuse plugs were observed. The Vshaped splits on the deflated tires are consistent with overload failure at impact. All damage to the landing gear appeared consistent with the gear being down at impact. The Safety Board found no evidence of malfunction of these systems. 2.2.2 Flight Controls The flap handle in the cockpit was found in the 23 degree position; however, there were no witness marks to indicate its position at impact. The cockpit tumbled during the accident sequence; therefore it is possible that the flap handle changed position. Also, the flap actuators did not contain witness marks and therefore were not conclusive in determining flap position. However, the inboard flap lockout cylinder was found with a witness mark that corresponded to a flap position of 12 degrees at impact. In addition, the CVR recorded the first officer stating that the flaps were set at 12 degrees. Therefore, it is reasonable to assume that the flaps were correctly set to 12 degrees for takeoff. An attempt was made to determine the rudder trim setting for takeoff. The rudder trim dial was found in a position corresponding to three units nose left trim.

[ANALYSIS] The three flight crewmembers were fatally injured. Visual meteorological conditions prevailed, and an instrument flight rules flight plan was filed. The flight was being conducted as a a ferry flight under Title 14 Code of Federal Regulations Part 91. The National Transportation Safety Board determines that the probable causes of this accident were: (1) the loss of directional control by the pilot in command during the takeoff roll, and his decision to continue the takeoff and initiate a rotation below the computed rotation airspeed, resulting in a premature liftoff, further loss of control and collision with the terrain. . (2) the flightcrew's lack of understanding of the three-engine takeoff procedures, and their decision to modify those procedures. (3) the failure of the company to ensure that the flightcrew had adequate experience, training, and rest to conduct the nonroutine flight. Contributing to the accident was the inadequacy of Federal Aviation Administration oversight of Air Transport International and Federal Aviation Administration flight and duty time regulations that permitted a substantially reduced flightcrew rest period when conducting a nonrevenue ferry flight under 14 Code of Federal Regulations Part 91. Safety issues discussed in the report focused on three-engine takeoff training and procedures, flightcrew fatigue, company crew assignment decisionmaking, and Federal Aviation Administration oversight of the company. Safety recommendations concerning these issues were made to the Federal Aviation Administration and Air Transport International. Also, as a result of the investigation of this accident, on March 30, 1995, the Safety Board issued Urgent Action Safety Recommendations A-95-38 and -39 to the Federal Aviation Administration concerning practices at Air Transport International. Fy NATIONAL TRANSPORTATION SAFETY BOARD WASHINGTON, D.C. 20594 AIRCRAFT ACCIDENT REPORT UNCONTROLLED COLLISION WITH TERRAIN AIR TRANSPORT INTERNATIONAL I DOUGLAS DC-8-63, N782AL KANSAS CITY INTERNATIONAL AIRPORT _ KANSAS CITY, MISSOURI » FEBRUARY 16, 1995 1, FACTUAL INFORMATION 1.1 History of Flight On Thursday, February 16, 1995, at 2027 CST', a Douglas DC-8-63, N782AL, operated by Air Transport International (ATD, was destroyed by ground impact and fire during an attempted takeoff at the Kansas City International Airport (MCI), Kansas City, Missouri. The three flight crewmembers were fatally injured. Visual meteorological conditions prevailed, and an instrument flight rules (IFR) flight plan was filed.

[ANALYSIS] The chief pilot telephoned the captain prior to the ferry flight and discussed a possible crosswind problem at the destination airport and the matter of a landing curfew there. He did not, however, review three-engine takeoff procedures with him. The Safety Board believes that had the takeoff been discussed in more detail, it might have become apparent to the chief pilot that the captain did not fully comprehend the three-engine takeoff procedure. During the investigation, a survey of nine other cargo operators revealed that only two used line flightcrews for three-engine takeoffs, and that one of those two operators restricted three-engine takeoffs to only "the most experienced and selected” flightcrews. Seven of the nine restrict such takeoffs to only management flightcrews, such as check airmen or special maintenance ferry crews. Therefore, the Safety Board concludes ATI’s policy of routinely assigning line 63 flightcrews for such operations, when almost all other operators restrict such flights, must be considered inappropriate. 2.4 Flightcrew Performance 2.4.1 Engine Start The engine start sequence was interrupted because the flightcrew did not ensure that all appropriate circuit breakers were in on the No. 4 engine. While attempting to start this engine, it was obvious that the captain was unfamiliar with the starter duty cycle limitations, and he did not determine the correct limitations by reference to the flight manual. The flight engineer called attention to the matter during multiple start attempts of this engine. 2.4.2 Landing Curfew The Safety Board believes that the flightcrew was concerned about trying to reach their destination before the landing curfew at Westover Airport, and that the crewmembers were unaware that the curfew time could be extended through ATI management channels. Prior to taxiing, the captain said that they should try to fly direct routes between navigational aids, in order to reduce the en route flight time. After the first takeoff attempt, the flightcrew again discussed the subject of trying to reach the destination airport. The comments by the first officer, "boy it's gettin’ tight," followed by, "hey we did our best you know," clearly indicated continued concern over the curfew and their desire to arrive before the airport closed. In addition, a time and distance calculation revealed that following the turn off the runway after the rejected takeoff, the flightcrew taxied the airplane to the departure end of the runway for another attempt at an average taxi speed of about 26 knots (about 30 miles per hour). The Safety Board believes that this is at, or may even exceed, the limit for a safe taxi speed, especially at night, and during a time when all three crewmembers were talking about the previous rejected takeoff. Therefore, the Safety Board believes that the flightcrew was convinced that they should arrive at their destination prior to the landing curfew, and that they were preoccupied with this goal.

[ANALYSIS] This increased his mental workload dramatically and probably contributed directly to the accident. The flight engineer could have placed himself in a similar predicament to that of the captain, if he was adding power on the asymmetric engine in response to the directional control inputs of the captain. Lastly, if the captain believed there was any possibility that a mechanical engine acceleration problem existed, the Safety Board finds it difficult to explain why he relinquished control of the throttle to another crewmember. 67 Shortly after the captain agreed to the unconventional takeoff procedure, the flight engineer asked the captain, "how much rudder were you stickin' in?" The captain replied, "I had it all the way in." This fact should have triggered a thorough, deliberate examination of all facets of the aborted takeoff, including a recalculation of Vmcg. However, there was never a discussion about why directional control could not be maintained, even though the captain used all the available rudder. Shortly thereafter, the subject of the power increase again came up, when the captain said, "it seemed what happened, it was goin' up smoothly and then all of a sudden...it jerked and then yeah." The first officer then made a statement which clearly indicated that he did not understand the concept of Vmcg. The first officer said, "...when we...get near Vmcg or get near Vr or Vmcg if we're usin’ all our rudder authority you might wanta' consider abort possibly because once we get higher we're gunnar be...in even worse trouble correct." The captain replied, "that's correct absolutely." The flight engineer challenged the statement by saying, "No actually above Vmcg you[r] rudder has more authority it's helping you more." The captain did not respond to this statement, which was, in fact, correct. The flight engineer went on to describe a four-engine takeoff with the loss of an engine by stating, "if we were to lose ah about the time an outboard engine before Vmcg...you can't control the takeoff because you will lose directional control because you[r] other engine is already in." This statement, although correct, may have further confused the captain and the first officer, because it was not clear that he was describing a four-engine takeoff, rather than the takeoff at hand. The first officer then said, "okay yeah you're right you're one hundred percent right." The captain was silent at that point. The Safety Board believes that the only person in the cockpit who had an understanding of the basic concept of a three-engine takeoff was the flight engineer. It is not clear, however, if any of the flightcrew understood the concept of the V speeds as applied to the three-engine takeoff. The accident takeoff is compared to a Douglas demonstration of an ideal three-engine takeoff in figure 7. On the accident takeoff, the power on the No. 4 engine was increased at a more rapid rate than on the first takeoff.

[ANALYSIS] At this position, the flight crew would have been able to see the runway 26 holding position sign, the “26” painted runway number, the taxiway A lights across runway 26, and the runway 22 holding position sign in the distance.151 FDR data showed that, about 0605:24, the captain began to taxi the airplane across the runway 26 hold short line. FDR data also showed that, about 0605:41, the airplane began to turn onto runway 26, and the CVR showed that, about 0605:46, the first officer completed the lineup checklist. 150  In accordance with FAA Order 7110.65, “Air Traffic Control,” paragraph 3-7-2, the controller would have been required to provide turn-by-turn directions to the departure runway if the flight crew had so requested. 151  During the 50-second timeframe, the controller did not query the flight crew regarding why the airplane was stopped at the hold short line for runway 26. The controller’s actions during the taxi and attempted takeoff sequence are discussed in section 2.2.3. Analysis National Transportation Safety Board A I R C R A F T Accident Report 60 Takeoff Roll 2.2.1.3  About 0605:58, the captain transferred control of the airplane to the first officer, stating, “all yours,” to which the first officer acknowledged, “my brakes, my controls.” At this time, the captain would have switched his attention from outside to inside the cockpit, and the first officer would have switched his attention from inside to outside the cockpit. About 2 seconds later, the airplane was aligned with the centerline for runway 26. The CVR recording showed that the flight crew had referred to runway 22 as the departure runway multiple times before takeoff, and FDR data showed that the pilots’ heading bugs were set to 227º, which was consistent with the magnetic heading for runway 22. The Safety Board concludes that the captain and the first officer believed that the airplane was on runway 22 when they taxied onto runway 26 and initiated the takeoff roll. About 0606:16, the first officer stated, “[that] is weird with no lights,” to which the captain responded, “yeah.” At that time, the airplane was passing through the intersection of runway 26 with runway 22. About 0606:24, the captain called the 100‑knot airspeed check. About the same time, the airplane accelerated beyond the maximum airspeed that would have allowed the airplane to remain on the available runway if the flight crew had rejected the takeoff and used maximum braking.152 At 0606:31.2, the captain called, “V one, rotate,” followed immediately by his exclamation, “whoa.” The aircraft performance study for this accident showed that, at the time of the VR callout, the airplane was 236 feet from the end of the runway.

[ANALYSIS] If the controller had noticed that the airplane had stopped in that location before issuing the takeoff clearance, the controller could have queried the flight crew, issued additional taxi instructions, or closely monitored the airplane’s subsequent progress. The controller’s postaccident statements indicated that he did not notice that the airplane had stopped short of runway 26. As a result, he missed an opportunity to prevent the flight crew’s surface navigation error and subsequent wrong runway takeoff attempt. The second time period lasted 28 seconds. It began about 0605:56, when the airplane began to align with runway 26, and ended about 0606:24, when the airplane accelerated beyond the maximum airspeed that would have allowed the airplane to remain on the available runway if the flight crew rejected the takeoff and used maximum braking. The airplane’s movements during this time were not consistent with the clearance provided by the controller and were a clear sign of a lack of positional awareness on the part of the flight crew. If the controller had been looking out the tower cab window and monitoring the flight, he could have addressed this situation by alerting the flight crewmembers that the airplane was on the wrong runway or, later, by instructing them to reject the takeoff. The controller did not take any such actions. The controller indicated that he did not see the airplane align with runway 26 or begin its takeoff roll because he had turned around to perform the traffic count, which is an administrative record-keeping task. The Safety Board examined possible reasons why the controller did not notice indications of the flight crew’s surface navigation error either during the 50-second window of opportunity, which began about 2 minutes before the accident, or during the 28-second critical window, which began about 39 seconds before the accident. Figure 7 shows a detailed timeline of communications and events that occurred in the 2 minutes before the accident. 192  Flight crews stop along a taxi route for various reasons, including a perception of potential traffic conflicts, passenger movements in the cabin, and uncertainty about the taxi clearance or the taxi routes. Some stops during taxi are accompanied by radio transmissions to ATC to explain the delay or seek clarification. Some controllers have a heightened sense of awareness when stops are made during taxi depending on the circumstances and the duration. Analysis National Transportation Safety Board A I R C R A F T Accident Report 78 Air Traffic Control Event Timeline Figure 7.  Note: COM, Comair; EGF, American Eagle; SKW, SkyWest; ZID, Indianapolis Air Route Traffic Control Center; TMU, traffic management unit. Analysis National Transportation Safety Board A I R C R A F T Accident Report 79 Window of Opportunity During Which the Airplane Was Stopped 2.2.3.1  Short of the Wrong Runway On the day of the accident, all LEX tower and radar positions were combined and were being worked by one controller.193 As a result, the controller had to switch his attention between tower and radar tasks.

[FINDINGS] 103 Conclusions 3. Findings 3.1  The captain and the first officer were properly certificated and qualified under Federal 1. regulations. There was no evidence of any medical or behavioral conditions that might have adversely affected their performance during the accident flight. Before reporting for the accident flight, the flight crewmembers had rest periods that were longer than those required by Federal regulations and company policy. The accident airplane was properly certified, equipped, and maintained in accordance 2. with Federal regulations. The recovered components showed no evidence of any structural, engine, or system failures. Weather was not a factor in this accident. No restrictions to visibility occurred during 3. the airplane’s taxi to the runway and the attempted takeoff. The taxi and the attempted takeoff occurred about 1 hour before sunrise during night visual meteorological conditions and with no illumination from the moon. The captain and the first officer believed that the airplane was on runway 22 when 4. they taxied onto runway 26 and initiated the takeoff roll. The flight crew recognized that something was wrong with the takeoff beyond the 5. point from which the airplane could be stopped on the remaining available runway. Because the accident airplane had taxied onto and taken off from runway 26 without 6. a clearance to do so, this accident was a runway incursion. Adequate cues existed on the airport surface and available resources were present in 7. the cockpit to allow the flight crew to successfully navigate from the air carrier ramp to the runway 22 threshold. The flight crewmembers’ nonpertinent conversation during the taxi, which was not in 8. compliance with Federal regulations and company policy, likely contributed to their loss of positional awareness. The flight crewmembers failed to recognize that they were initiating a takeoff on the 9. wrong runway because they did not cross-check and confirm the airplane’s position on the runway before takeoff and they were likely influenced by confirmation bias. Even though the flight crewmembers made some errors during their preflight activities 10. and the taxi to the runway, there was insufficient evidence to determine whether fatigue affected their performance. Conclusions National Transportation Safety Board A I R C R A F T Accident Report 104 The flight crew’s noncompliance with standard operating procedures, including the 11. captain’s abbreviated taxi briefing and both pilots’ nonpertinent conversation, most likely created an atmosphere in the cockpit that enabled the crew’s errors. The controller did not notice that the flight crew had stopped the airplane short of the 12. wrong runway because he did not anticipate any problems with the airplane’s taxi to the correct runway and thus was paying more attention to his radar responsibilities than his tower responsibilities. The controller did not detect the flight crew’s attempt to take off on the wrong runway 13. because, instead of monitoring the airplane’s departure, he performed a lower-priority administrative task that could have waited until he transferred responsibility for the airplane to the next air traffic control facility.

[ANALYSIS] Even though discrepancies existed between the airport chart and the external cues available to the pilots because of an ongoing construction project at the airport,157 the chart depicted the paved taxiway and runway surfaces at the time of the accident. Another available resource within the cockpit was the instrumentation, including the heading bug, which had been set to 227º to correspond to the magnetic heading for runway 22. This heading information, which was clearly presented on both flight crewmembers’ MFDs and PFDs, would have provided the pilots with a real-time cue of their orientation relative to runway 22. 156  Although the first officer’s attention was focused inside the cockpit while the airplane was taxiing to the departure runway, he would have had multiple opportunities to look outside the cockpit and monitor the airplane’s progress along the taxi route. Such monitoring would have helped the first officer gauge and pace his activities with the available time left for taxi. 157  Taxiway A north of runway 26 had been closed (as indicated by a local NOTAM) and barricaded, and taxiway A5 had been redesignated as taxiway A. Neither of these changes was depicted on the airport chart. However, no evidence indicated that either pilot was confused by the discrepancies. In addition, the pilots of SkyWest flight 6819 and American Eagle flight 882 used the same chart to navigate to runway 22. Analysis National Transportation Safety Board A I R C R A F T Accident Report 63 In addition, the flight crewmembers could have communicated with the controller if they became unsure of their position at any time.158 However, the CVR did not record any statement by either flight crewmember about being unsure of the airplane’s position on the airport surface at any time during the taxi and takeoff roll. The CVR also did not record any indication that the crewmembers had attempted to confirm the airplane’s position on the runway before beginning the takeoff roll. The taxi routing and cues available to the accident pilots were identical to those that were available to the pilots of the two regional jets (SkyWest flight 6819 and American Eagle flight 882) that departed before Comair flight 5191. The SkyWest and American Eagle pilots had no difficulty identifying, and successfully navigating to, runway 22 using the available cues,159 even with the differences in taxiway signage and chart labeling.160 In addition, even though the airport configuration at the time of the accident had been in place for 1 week, the Safety Board did not identify any reports about surface navigation problems at LEX during that time.161 The Safety Board concludes that adequate cues existed on the airport surface and available resources were present in the cockpit to allow the flight crew to successfully navigate from the air carrier ramp to the runway 22 threshold. Preflight Activities and Actions During the Taxi 2.2.2.2  Because the availability of cues and aids for the pilots’ wayfinding task was not a factor in this accident, the Safety Board examined the crew’s actions during the preflight and taxi phases of the flight’s operation to identify possible reasons for the error.

[CONCLUSIONS] Because of an ongoing construction project at Blue Grass Airport, the taxiway identifiers 25. represented in the airport chart available to the flight crew were inaccurate, and the information contained in a local notice to airmen about the closure of taxiway A was not made available to the crew via automatic terminal information service broadcast or the flight release paperwork. The controller’s failure to ensure that the flight crew was aware of the altered taxiway A 26. configuration was likely not a factor in the crew’s inability to navigate to the correct runway. Because the information in the local notice to airmen (NOTAM) about the altered 27. taxiway A configuration was not needed for the pilots’ wayfinding task, the absence of the local NOTAM from the flight release paperwork was not a factor in this accident. The presence of the extended taxiway centerline to taxiway A north of runway 8/26 28. was not a factor in this accident. Probable Cause 3.2  The National Transportation Safety Board determines that the probable cause of this accident was the flight crewmembers’ failure to use available cues and aids to identify the airplane’s location on the airport surface during taxi and their failure to cross‑check and verify that the airplane was on the correct runway before takeoff. Contributing to the accident were the flight crew’s nonpertinent conversation during taxi, which resulted in a loss of positional awareness, and the Federal Aviation Administration’s failure to require that all runway crossings be authorized only by specific air traffic control clearances. National Transportation Safety Board A I R C R A F T Accident Report 106 Recommendations 4. New Recommendations 4.1  As a result of the investigation of this accident, the National Transportation Safety Board makes the following recommendations: —To the Federal Aviation Administration: Require that all 14 Code of Federal Regulations Part 91K, 121, and 135 operators establish procedures requiring all crewmembers on the flight deck to positively confirm and cross-check the airplane’s location at the assigned departure runway before crossing the hold short line for takeoff. This required guidance should be consistent with the guidance in Advisory Circular 120‑74A and Safety Alert for Operators 06013 and 07003. (A-07-44) Require that all 14 Code of Federal Regulations Part 91K, 121, and 135 operators install on their aircraft cockpit moving map displays or an automatic system that alerts pilots when a takeoff is attempted on a taxiway or a runway other than the one intended. (A-07-45) Require that all airports certificated under 14 Code of Federal Regulations Part 139 implement enhanced taxiway centerline markings and surface painted holding position signs at all runway entrances. (A-07-46) Prohibit the issuance of a takeoff clearance during an airplane’s taxi to its departure runway until after the airplane has crossed all intersecting runways. (A-07-47) Revise Federal Aviation Administration Order 7110.65, “Air Traffic Control,” to indicate that controllers should refrain from performing administrative tasks, such as the traffic count, when moving aircraft are in the controller’s area of responsibility. (A-07-48)

[CONCLUSIONS] The controller did not detect the flight crew’s attempt to take off on the wrong runway 13. because, instead of monitoring the airplane’s departure, he performed a lower-priority administrative task that could have waited until he transferred responsibility for the airplane to the next air traffic control facility. The controller was most likely fatigued at the time of the accident, but the extent 14. that fatigue affected his decision not to monitor the airplane’s departure could not be determined in part because his routine practices did not consistently include the monitoring of takeoffs. The Federal Aviation Administration’s operational policies and procedures at the 15. time of the accident were deficient because they did not promote optimal controller monitoring of aircraft surface operations. The first officer’s survival was directly attributable to the prompt arrival of the first 16. responders; their ability to extricate him from the cockpit wreckage; and his rapid transport to the hospital, where he received immediate treatment. The emergency response for this accident was timely and well coordinated. 17. A standard procedure requiring 14 18. Code of Federal Regulations Part 91K, 121, and 135 pilots to confirm and cross-check that their airplane is positioned at the correct runway before crossing the hold short line and initiating a takeoff would help to improve the pilots’ positional awareness during surface operations. The implementation of cockpit moving map displays or cockpit runway alerting 19. systems on air carrier aircraft would enhance flight safety by providing pilots with improved positional awareness during surface navigation. Enhanced taxiway centerline markings and surface painted holding position 20. signs provide pilots with additional awareness about the runway and taxiway environment. This accident demonstrates that 14 21. Code of Federal Regulations 91.129(i) might result in mistakes that have catastrophic consequences because the regulation allows an airplane to cross a runway during taxi without a pilot request for a specific clearance to do so. Conclusions National Transportation Safety Board A I R C R A F T Accident Report 105 If controllers were required to delay a takeoff clearance until confirming that an 22. airplane has crossed all intersecting runways to a departure runway, the increased monitoring of the flight crew’s surface navigation would reduce the likelihood of wrong runway takeoff events. If controllers were to focus on monitoring tasks instead of administrative tasks when 23. aircraft are in the controller’s area of operations, the additional monitoring would increase the probability of detecting flight crew errors. Even though the air traffic manager’s decision to staff midnight shifts at Blue Grass 24. Airport with one controller was contrary to Federal Aviation Administration verbal guidance indicating that two controllers were needed, it cannot be determined if this decision contributed to the circumstances of this accident. Because of an ongoing construction project at Blue Grass Airport, the taxiway identifiers 25. represented in the airport chart available to the flight crew were inaccurate, and the information contained in a local notice to airmen about the closure of taxiway A was not made available to the crew via automatic terminal information service broadcast or the flight release paperwork.

[ANALYSIS] Because no other traffic was on the airport surface to pose a conflict to the airplane during its taxi, the controller would not have expected much useful information to be obtained by frequently scanning the runway environment, which would have decreased the likelihood that he was frequently looking out the tower cab windows at the runway environment during the 50-second window of opportunity. The Safety Board concludes that the controller did not notice that the flight crew had stopped the airplane short of the wrong runway because he did not anticipate any problems with the airplane’s taxi to the correct runway and thus was paying more attention to his radar responsibilities than his tower responsibilities. Postaccident observations from the LEX ATCT revealed that, from the controller’s work station at night, it was somewhat difficult to see whether the CRJ-100 demonstration airplane was located at the hold short line for runway 26, on taxiway A, or at the hold short line for runway 22 because of (1) the proximity of these locations in the controller’s visual field as a result of the new taxiway configuration198 and (2) the reduced visibility of ground texture, linear perspective, and other monocular depth cues that are useful for judging distances beyond 15 to 20 feet. Although the controller had been working at the tower for 17 years and had presumably become an expert at recognizing aircraft positions on the airport surface, the use of former taxiway A5 (redesignated as taxiway A) to reach the runway 22 threshold was new to him because the runway had been shifted and the redesignated taxiway had been in place for only 1 week (four of the controller’s shifts) 198  From the tower cab, the hold short lines for runways 22 and 26 appeared close to each other because they were separated by less than 5º of visual angle. Analysis National Transportation Safety Board A I R C R A F T Accident Report 81 at the time of the accident.199 These factors would have made it more difficult for the controller to determine the airplane’s exact location. In addition, the controller was not required to determine that the airplane had reached the departure runway before he cleared the airplane for takeoff; he was only expected to determine that the airplane was at a location that was consistent with its taxi clearance. When the flight crew had stopped the airplane at the runway 26 hold short line, the airplane was in a location that was consistent with its taxi clearance. The controller reported that he did not see the airplane stop in this position. Even if the controller had seen the airplane at that time and noticed that it was not moving, a brief scan of the runway environment would not have informed him of whether the airplane had been stopped only briefly or for a longer period of time. Nevertheless, the controller could have detected that the airplane had stopped short of the wrong runway if he had been monitoring the airplane’s progress along the taxi route.

[ANALYSIS] The appearance of the runway end environment would have provided a salient cue to the flight crew that the airplane was in an extremely hazardous situation and could not remain on the ground. The airplane’s airspeed at the time of the captain’s VR callout was 131 knots, which was 11 knots below the planned VR airspeed of 142 knots (which the flight crew had briefed and entered into the airplane’s EFIS).153 Thus, the captain’s early VR callout and subsequent “whoa” exclamation indicated that he recognized that something was wrong with the takeoff. FDR data showed that, in response, the first officer pulled the control column full aft154 and that the airplane rotated at a rate of about 10º per second, which is three times the normal rotation rate. This abnormal column input showed that the first officer also recognized that something was wrong with the takeoff. Although numerous cues, including the lack of runway lighting, were available to indicate that the airplane was not on the assigned runway (see sections 2.2.2.1 and 2.2.2.3), 152  According to calculations by Bombardier, for the airplane to have stopped before the end of runway 26, maximum braking would have had to start when the airplane was at an airspeed of about 103 knots. 153  FDR data for the accident airplane’s 12 previous takeoffs indicated that rotation occurred at or after reaching the VR airspeed. 154  FDR data showed that the left and right control column inputs during the accident rotation reached 10.6º and 10.9º, respectively, and that the nominal control column input for rotation during the accident airplane’s 12 previous takeoffs was between 4º and 5º. Analysis National Transportation Safety Board A I R C R A F T Accident Report 61 the flight crew had not correctly interpreted these cues or noticed them until after it was too late to successfully abort the takeoff. The Safety Board concludes that the flight crew recognized that something was wrong with the takeoff beyond the point from which the airplane could be stopped on the remaining available runway.

[ANALYSIS] Analysis National Transportation Safety Board A I R C R A F T Accident Report 59 instruction. Because two airplanes, SkyWest flight 6819 and American Eagle flight 882, were given the same taxi clearance and had already correctly taxied to and held short of runway  22 without any special instructions, there was no apparent reason for the controller to have suspected that the pilots would have had difficulty navigating to the departure runway.150 Taxi to Runway 2.2.1.2  From about 0603:16 to 0603:56, while the captain was taxiing the airplane and performing navigational checking activities, both pilots resumed the nonpertinent discussion that was started while the airplane was parked at the gate. (Figure 1 shows the location of the airplane along the taxi route while this conversation was occurring.) The nonpertinent conversation was not in compliance with the sterile cockpit rule required by company procedures and 14 CFR 121.542 (see section 1.17.1.3). The primary reason for the sterile cockpit rule is to ensure that the pilots’ attention is directed to operational concerns during critical phases of flight (including taxi) and is not redirected or degraded because of nonessential activities or discussion. FDR data showed that, about 0604:33, the airplane stopped on taxiway A at the hold short line for runway 26, which was about 560 feet from the intended destination— the hold short line for runway 22. During this time, the first officer was completing the before takeoff checklist. About 0605:15, the first officer advised the controller that the airplane was ready to depart, and the controller told the flight crew that the airplane should fly the runway heading and was cleared for takeoff. Neither the first officer nor the controller stated the runway number during the request and clearance for takeoff, but ATC procedures did not require them to do so. Because the flight crew believed that the airplane was at the hold short line for runway 22 at the time of the takeoff clearance (see section 2.2.2.3), the absence of a reference to runway 22 in the request and clearance for takeoff was not a factor in this accident. The 50-second timeframe during which the airplane was stopped at the runway 26 hold short line should have provided the flight crew with ample time to look outside the cockpit and determine the airplane’s position on the airport. At this position, the flight crew would have been able to see the runway 26 holding position sign, the “26” painted runway number, the taxiway A lights across runway 26, and the runway 22 holding position sign in the distance.151 FDR data showed that, about 0605:24, the captain began to taxi the airplane across the runway 26 hold short line.

[ANALYSIS] Analysis National Transportation Safety Board A I R C R A F T Accident Report 67 At the time that the first officer began to increase thrust for takeoff, FDR data showed that the magnetic heading of the airplane was about 266º, which corresponded to the magnetic heading for runway 26. Figures 6 and 6a show the approximate configuration of the captain’s MFD and PFD, including the heading bug setting and magnetic heading indication, when the airplane was lined up on the runway 26 centerline. As shown in the figures, the heading bug was offset by 40º, providing a salient cue that the airplane was not lined up on the correct departure runway. The CVR did not record any awareness by the flight crewmembers about this offset. Multifunction Display Figure 6.  Source: Rockwell Collins. The display was configured as requested by the Safety Board. Analysis National Transportation Safety Board A I R C R A F T Accident Report 68 Primary Flight Display Figure 6a. Source: Rockwell Collins. The display was configured as requested by the Safety Board. The wayfinding task includes an ongoing cross-check between an airplane’s expected and actual position using available cues in the environment and aids in the cockpit. The CVR did not record any discussion by the flight crew about the need to cross-check the airplane’s position on the runway.171 The Safety Board concludes that the flight crewmembers failed to recognize that they were initiating a takeoff on the wrong runway because they did not cross-check and confirm the airplane’s position on the runway before takeoff and they were likely influenced by confirmation bias. 171  On March 23, 2007, Comair revised its operations manual to include a departure runway checklist item. Analysis National Transportation Safety Board A I R C R A F T Accident Report 69 Fatigue 2.2.2.4  The Safety Board examined whether pilot fatigue could have been a factor in this accident by assessing the preconditions that could allow for the development of a fatigued state and examining the nature of the pilots’ demonstrated performance deficiencies. Potential conditions that can lead to the development of a fatigued state include chronic sleep restriction, acute sleep loss, circadian disruption (work during times when one would normally be asleep), and time since awakening. The captain and the first officer received more than the minimum required rest periods during their respective trips in the days before the accident, and their flight and duty times in the week and month before the accident would not have precluded them from obtaining adequate sleep.

[ANALYSIS] Dismukes, “Concurrent Task Management and Prospective Memory: Pilot Error as a Model for the Vulnerability of Experts.” Proceedings of the Human Factors and Ergonomics Society 50th Annual Meeting, 2006. 164  R.K. Dismukes, G.E. Young, and R.L. Sumwalt, Cockpit Interruptions and Distractions: Effective Management Requires a Careful Balancing Act. ASRS Directline, vol. 10, pages 4-9, 1998. Analysis National Transportation Safety Board A I R C R A F T Accident Report 65 in either a slight deviation from an intended path, or becoming totally lost.”165 In addition, active attention is thought to be necessary for maintaining situational awareness, and the allocation of attention to irrelevant stimuli can degrade awareness.166 Finally, research on distractions while driving found that drivers are less likely to detect changes that have occurred in the environment when they are involved in casual conversation because being engaged in a conversation may degrade the encoding of visual information.167 The first officer initiated the nonpertinent conversation as the captain was navigating along the taxi route. The captain had the responsibility to assert both his leadership role and command authority to stop the discussion. Rather, the captain allowed the conversation to continue and participated in it. Also, instead of initiating the nonpertinent conversation, the first officer should have been monitoring the captain’s actions and independently assessing the airplane’s location along the taxi route. The Safety Board concludes that the flight crewmembers’ nonpertinent conversation during the taxi, which was not in compliance with Federal regulations and company policy, likely contributed to their loss of positional awareness. It is important to note that the CVR did not record any statement by either flight crewmember about this loss of positional awareness. Cues to Indicate a Takeoff From Runway 22 2.2.2.3  The presence of runway markings—a white centerline and side stripes—ahead of the airplane would have facilitated the captain’s perception that the airplane had arrived at the hold short line for runway 22, even though the airplane was actually at the hold short line for runway 26. In addition, the angle from the runway 26 hold short line on taxiway A to runway 26 was the same as the angle from the runway 22 hold short line on former taxiway A (north of runway 8/26) to runway 22. Also, the taxiway A centerline split into three lines after the runway 26 hold short line.

[ANALYSIS] This analysis discusses the taxi and attempted takeoff sequence and the associated human factor issues, survival factors, efforts to mitigate surface navigation errors, ATC staffing, and other issues related to the accident flight. Taxi and Attempted Takeoff Sequence 2.2  The pilots attempted to take off from a different runway than the airplane had been cleared to use. Section 2.2.1 details the events leading to the attempted takeoff, section 2.2.2 considers the pilot human factors that might have played a role in this surface navigation error, and section 2.2.3 explains the ATC human factors that might have played a role. Wrong Runway Departure 2.2.1  Before Taxi Activities 2.2.1.1  The Comair Operations Manual indicated that the captain was to conduct the taxi briefing. The briefing was to incorporate Comair standard taxi information, including that both flight crewmembers should have the appropriate airport diagrams out and available and that traversing runways required extra diligence. Comair standard taxi information was to be briefed in its entirety for the first flight as a crew and could be abbreviated to “Comair standard” for subsequent flights. The accident flight was the pilots’ first flight Analysis National Transportation Safety Board A I R C R A F T Accident Report 58 as a crew, but, about 0556:14, the captain stated “Comair standard” instead of all of the information in that portion of the taxi briefing, including that runway 26 was to be crossed while navigating to runway 22. The abbreviated briefing was contrary to company policy, and the Safety Board notes that it is prudent for pilots to fully conduct taxi briefings according to standard operating procedures. However, despite this abbreviated briefing (which occurred about 10 minutes before the accident), multiple and more salient cues existed to aid the flight crew while navigating to the runway, and the airport navigation task was relatively simple,146 as discussed in this analysis. The first officer was the flying pilot for the accident flight. About 0557:23, he briefed the taxi route as part of the takeoff briefing,147 stating, “let’s take it out and … take … [taxiway] Alpha. Two two’s a short taxi.” Although the CVR did not record either pilot explicitly referencing the airport chart, this statement is consistent with the first officer examining the chart because no specific taxi instructions had been provided to the flight crew. Also, the number of times that each crewmember had previously arrived at or departed from LEX was likely not sufficient to allow either one to have memorized taxiway identifiers and routes.

[ANALYSIS] Analysis National Transportation Safety Board A I R C R A F T Accident Report 64 was included in the flight release paperwork, and started its APU. Although these actions likely consumed a portion of the crew’s available time at the gate, CVR evidence and interviews with ground personnel indicated that neither pilot appeared to be rushed or hurried as he completed required tasks. For example, as stated in section 2.2.1.2, the CVR recorded both crewmembers involved in a detailed nonpertinent discussion before and during pushback from the gate, which indicated that both crewmembers were relaxed. In addition, FDR data showed that the navigation to the runway was conducted at a normal pace, and the controller stated that the pilots did not seem to be rushed. Thus, the available evidence indicated that the flight crew was not under time pressure. The Safety Board reviewed the CVR recording to evaluate the flight crew’s workload and focus of attention during the taxi to the runway. While the captain was maneuvering the airplane along the taxi route, the first officer was performing preflight checklists, including some additional items that were necessary because the flight was the airplane’s first flight of the day. Although FDR data showed that the taxi lasted only about 2 1/2 minutes, the first officer had adequate time to complete the preflight activities. It is important to note that the first officer was experienced in his position and that these activities, including the first-flight-of-the-day items, would have been well learned and would not have presented him with a high workload condition. In addition to navigating the airplane along its taxi route, the captain would have been overseeing the first officer’s performance and providing cross-checking as necessary. The captain was also experienced in his position and would have been skilled at dividing his attention between controlling the airplane, navigating the airport, and monitoring the first officer’s performance. Moreover, with the captain’s level of experience, the control inputs associated with controlling an airplane on an airport surface would have required very little cognitive effort. The nonpertinent conversation occurred during 40 seconds of the 150-second taxi time (from about 0603:16 to about 0603:56). The timing of this discussion and its duration are the most salient evidence to demonstrate that neither pilot was experiencing high workload at the time or considered the taxi to runway 22 to be a challenging task. However, people have limited attention resources, and, when distracted by conversation, both real-time processing of information and prospective memory (that is, remembering to do something at a later time) can suffer.163 More than 20 years of research using the ASRS database has shown that pilots cite social conversation with other crewmembers as an activity that can distract the pilots from tasks that must be performed.164 Also, human factors research on aviation wayfinding indicated that any breakdown in a pilot’s assessment of position “can result 163  R.K.

[ANALYSIS] For example, in its report on the August 16, 1987, Northwest Airlines flight 255 accident, the Board found that checklists (except for the before start checklist) were not being called for or accomplished according to company procedures and that, after pushback, the flight crew initiated conversations that were not related to duty requirements and diverted the crew’s attention from task-related activities. Regarding the role of cockpit discipline in these errors, the report stated, “it is the captain’s responsibility to structure the manner in which his crew will accomplish its duties … he must set the tone for how this information will be proffered.” The report further stated that a well-defined role structure in the cockpit reduces ambiguity about each crewmember’s responsibilities and when those responsibilities will be performed.183 In addition, data from the LOSA Collaborative184 showed that flight crewmembers who intentionally deviated from standard operating procedures were three times more likely to commit other types of errors, mismanage more errors, and find themselves in more undesired aircraft situations compared with those flight crewmembers who did not intentionally deviate from procedures. The Safety Board concludes that the flight crew’s noncompliance with standard operating procedures, including the captain’s abbreviated taxi briefing and both pilots’ nonpertinent conversation, most likely created an atmosphere in the cockpit that enabled the crew’s errors. Summary of Pilot Human Factors 2.2.2.6  Even with the discrepancies that existed between airport charts and signage, the navigational task that the pilots of Comair flight 5191 faced—taxi to runway 22 via taxiway A—was straightforward and inherently simple. The primary cues for this navigational task—the airport markings and signage at hold short positions—were accurate and available to the flight crew. Flight crews are responsible for maintaining positive control of an airplane during all phases of flight, including surface operations. An essential component to positive control is knowing the position of the airplane at all times. It was the accident pilots’ responsibility, once they were cleared to taxi, to safely maneuver the airplane to runway 22. Each pilot had experience and expertise to allow the successful completion of this task using only the standard airport marking conventions; that is, once the pilots detected and recognized signs off the air carrier ramp that identified taxiway A, they 183  National Transportation Safety Board, Northwest Airlines, Inc., McDonnell Douglas DC-9-82, N312RC, Detroit Metropolitan Wayne County Airport, Romulus, Michigan, August 16, 1987, Aircraft Accident Report NTSB/AAR-88/05 (Washington, DC: NTSB, 1988). 184  According to its Web site, the LOSA Collaborative is a network of researchers, safety professionals, pilots, and airline representatives collaborating to provide, among other things, oversight and implementation of LOSA and a forum for information exchange regarding LOSA. The referenced data were provided to the Safety Board in January 2007.

[ANALYSIS] The referenced data were provided to the Safety Board in January 2007. Analysis National Transportation Safety Board A I R C R A F T Accident Report 74 had only to follow those signs and the taxiway centerline to the runway holding position sign for runway 22. Although this task is more difficult during night conditions, when surrounding features in the environment and horizon as well as other cues are not easily detectable, the critical features along the navigation route were internally lit or were illuminated by the airplane’s external lighting system. Both pilots were described to be in good health in the days before the accident.185 Postaccident toxicological testing for both pilots did not detect the presence of alcohol186 or any substances known to affect performance. Both pilots were required to wear corrective lenses during flight operations. Evidence indicated that the captain was wearing contact lenses during the accident flight, but the Safety Board could not determine whether the first officer was wearing corrective lenses during the flight. Even if the first officer had not been wearing corrective lenses, his ability to verify the airplane’s position during the taxi to runway 22 would most likely not have been affected.187 Evidence indicated that both pilots were able to read and interpret features inside the cockpit at intermediate distances and detect objects outside the cockpit at greater distances, and the pilots were described to have normal night vision. Observations in a CRJ-100 airplane after the accident demonstrated that no objects along the taxi route would have produced temporary flash blindness188 that could have impeded the pilots’ ability to detect signage and surface markings.189 As previously stated, numerous cues were available to the flight crew to indicate that the airplane was not at the position on the airport surface that it was supposed to be. For example, when the airplane stopped at the runway 26 hold short line instead of the runway 22 hold short line, the flight crew would have been able to see the runway 26 holding position sign and painted runway numbers, the continuation of taxiway A across runway 26, and the runway 22 holding position sign. Also, when the airplane was in 185  The first officer tested positive for a low level of pseudoephedrine. This drug, at low levels, is not considered to cause impairment. Performance-based side effects of pseudoephedrine are similar to those of caffeine. CAMI does not quantify this substance unless it is above therapeutic levels. The Safety Board was not able to determine why and when the medication was taken. 186  On August 26, 2006, at 1826, the first officer purchased food and two bottles of beer at the hotel restaurant. (Section 1.17.1.6 describes the FAA’s and Comair’s alcohol use policy.) According to the CAMI toxicology report, the first officer tested negative for ethanol.

[ANALYSIS] Two two’s a short taxi.” Although the CVR did not record either pilot explicitly referencing the airport chart, this statement is consistent with the first officer examining the chart because no specific taxi instructions had been provided to the flight crew. Also, the number of times that each crewmember had previously arrived at or departed from LEX was likely not sufficient to allow either one to have memorized taxiway identifiers and routes. During the takeoff briefing, the first officer stated that the runway end identifier lights were out and then commented, “came in the other night it was like … lights are out all over the place.” The first officer was referring to observations he made on a repositioning flight that landed on runway 22 about 0140 on the day before the accident. (The right runway edge lights after the intersection of runway 22 with runway 26 were out at that time.) The first officer did not brief that the taxi to runway 22 required crossing runway 26. The Safety Board was unable to determine why this information was not included in the first officer’s briefing. It is possible that the simplicity of the taxi and the use of only one taxiway might have led him to assume that it was unnecessary to include this additional information in his briefing. During postaccident interviews, other pilots indicated that they would brief this “short taxi”148 in a similar manner. No evidence indicated that the pilots were unaware of the need to cross runway 26 to arrive at runway 22. About 0602:01, the first officer notified the controller that the airplane was ready to taxi. The controller then instructed the flight crew to taxi the airplane to runway 22. Title 14 CFR 91.129(i), “Takeoff, Landing, Taxi Clearance,” permits pilots instructed to taxi to a point on the airport surface to cross all intersecting runways along the taxi route (without stopping) except for the assigned departure runway.149 Thus, the controller’s clearance for the airplane to taxi to runway 22 complied with 14 CFR 91.129(i), and the first officer’s response of “taxi two two” was an appropriate acknowledgment of the taxi 146  Comair’s high threat taxi procedures are used at those airports with an operating environment that presented “exceptional” hazards to safe taxi. However, these procedures were not in place for operations at LEX at the time of the accident. 147  As stated in section 1.1, the takeoff briefing is part of the before starting engines checklist. 148  Taxiway A is used for the taxi to both runways 4 and 22, but the taxi to runway 22 is significantly shorter than the taxi to runway 4, as shown on airport charts (see appendix C). 149  This regulation is discussed further in section 2.4.4. Analysis National Transportation Safety Board A I R C R A F T Accident Report 59 instruction.

[ANALYSIS] The Safety Board was not able to determine what effect, if any, this information would have had on the circumstances of this accident. However, this possible cue would not have had the same salience as the primary cues— the airport markings and signage—that were accurate and available to the flight crew. In addition, the Board’s analysis of the accident determined that the taxi to runway 22 was relatively simple and could have been successfully conducted by using the cues and aids available to the flight crew. Thus, the Safety Board concludes that, because the information in the local NOTAM about the altered taxiway A configuration was not needed for the pilots’ wayfinding task, the absence of the local NOTAM from the flight release paperwork was not a factor in this accident. As stated in section 1.18.6, the FAA has planned initiatives to modernize the current NOTAM system. These initiatives include aligning the U.S. NOTAM system with that of ICAO by October 2007 so that U.S. NOTAM information can be processed and provided to flight crews in a more timely, accurate, complete, and traceable manner and having digital NOTAM data displayed in the cockpit in textual and graphical formats. For this accident, even though the information in the local NOTAM was not necessary for the flight crew’s successful navigation to the correct runway, such information might be necessary under different circumstances. Thus, the Safety Board is pleased that the FAA is taking a proactive role to improve the current NOTAM system. Presence of Extended Taxiway Centerline 2.6.4  One of the three centerlines that diverged from the taxiway A hold short position for runway 8/26 led to taxiway A north of runway 8/26, which had been closed and was blocked by low-profile barricades with flashing red lights. This extended taxiway centerline had not been removed at the time of the accident. Analysis National Transportation Safety Board A I R C R A F T Accident Report 102 The LEX construction plan called for the removal of taxiway A north of runway 8/26 and the associated extended taxiway centerline within 30 to 45 days of that taxiway’s closure. Because proper barricades had been put into place until such time, and because the CVR did not record any confusion between the flight crewmembers about the taxiway A configuration, the Safety Board concludes that the presence of the extended taxiway centerline to taxiway A north of runway 8/26 was not a factor in this accident. National Transportation Safety Board A I R C R A F T Accident Report 103 Conclusions 3. Findings 3.1  The captain and the first officer were properly certificated and qualified under Federal 1. regulations. There was no evidence of any medical or behavioral conditions that might have adversely affected their performance during the accident flight. Before reporting for the accident flight, the flight crewmembers had rest periods that were longer than those required by Federal regulations and company policy.

[ANALYSIS] Even if the controller had seen the airplane at that time and noticed that it was not moving, a brief scan of the runway environment would not have informed him of whether the airplane had been stopped only briefly or for a longer period of time. Nevertheless, the controller could have detected that the airplane had stopped short of the wrong runway if he had been monitoring the airplane’s progress along the taxi route. Critical Window During Which an Administrative Task Was Performed 2.2.3.2  After the LEX controller cleared Comair flight 5191 for takeoff, he told American Eagle flight 882 to contact the Indianapolis ARTCC. According to the ATC transcript, the handoff of the American Eagle flight to the Indianapolis ARTCC occurred about 0605:40. About that time, Comair flight 5191, the only airplane for which the controller had responsibility, was turning onto runway 26 and had not yet deviated from the issued taxi clearance. The controller’s next active task would be to establish contact with the Comair flight and provide departure services (radar tasks), but he would likely not have expected to perform this task for about 1 minute.200 This 1-minute pause in active control tasks afforded the controller greater flexibility in terms of his allocation of attention. The controller stated that, after the handoff of the American Eagle flight to the Indianapolis ARTCC, he began the combined traffic count, which was an administrative record-keeping task.201 The standard operating procedure at LEX was to perform the traffic count on an hourly basis. However, the controller stated that he normally accumulated flight progress strips throughout the night and performed the traffic count once toward the end of his shift. The controller estimated that the traffic count would take 2 to 5  minutes to complete.202 The controller was expecting to be relieved by the incoming day shift controller at 0630, so he most likely wanted to complete the traffic count by that time. However, 199  When taxiway A north of runway 8/26 was used to reach the runway 22 threshold, the hold short lines for runways 22 and 26 were farther apart in the controller’s visual field. 200  The time between flight crew acknowledgment of the takeoff clearance and the controller’s acknowledgment of radar contact was 41 and 78 seconds for the SkyWest and American Eagle departures, respectively. 201  When the tower was staffed with more than one controller, the controller at the radar data position recorded the radar traffic count, and the controller at the clearance delivery position recorded the tower traffic count. When the midnight shift was staffed with one controller, that controller performed a combined radar and tower traffic count. 202  As a result, the controller must have expected that he would have to interrupt this administrative task to provide radar services to the Comair flight.

[ANALYSIS] During the Safety Board's performance study, go-around maneuvers from 20° bank angles were simulated at altitude by simultane power, applying aft elevator control to prevent conta ¢roundplane, and applying full opposite aileron to level the airplane. On go-sround &ttempts at speeds above stickshaker activation (approximately 1.97 Vs or greater), AO problems were encountered. However, during similar go-around attempts at speeds within the shaker range, the downgoing wing was observed to Stall and resulted in nearly vertical bank angles, In this accident, 1.07 Vs was about 94 kn alld it is unlikely that the Learjet slowed to that Speed before the first roll, However, if leading edge ice accumulation had increased the Stall entry speed by 10 to 12 kn of greater, the equivalent 1.07 V3 air flow pattern over the wings may have been about 106 kn, It is possible the Learjet may have slowed to this airspeed and the go-around attempt in this comparative Speed range caused the aircraft to beyin the rolling maneuver, which resulted in a loss of altitude and ground impact. Such an occurrence would be consistent with the findings of the performance study, PHM ee ee However, the Safety Board could not determine the airspeed or acceleration of the Learjet as the go-around attempt was beg:n, and, therefore, cannot draw a eonelusion in this regard, The Safety Board found during flight performance studies at altitude that after wing rolloff following the stall, lateral control was not effective. The roll could be reversecdt with aileron and rudder, but at such a rate that 500 ft of altitude was lost during the reversal. After the stall, there was no tendency for the aircraft to abruptly reverse roll direction. The roll control power of the aireraft was good until stall entry; however, roll damping was low and control wheel centering due to force feedback was low. The Icw roll control sensitivity was due to the large 110° wheel throw and a reiease of roll input did not stop the roll rate. These handling characteristics are different from lower performance aircraft and could account for response actions in which some pilots might tend to overcontrol if they used sudden, large and untimely control wheel displacements. In this accident, low roll control sensitivity and the possibility of a sudden pilot reaction might have sustained the rolling maneuvers following the initial wing drop, The influence of ground effect on the Learjet Century III's lateral contro] during the rolling maneuvers is not known and the Safety Board was not able to explore the potential influence of ground effect at large bank angles because of the obvious risks involved in flight testing under these conditions, However, as ground effect increases wing lift, induced drag decreases and the total effeet on the wing may increase roll contro! authority near stall airspeed. The influence of ground effect on a stalled wing rolling downward, however, would probably be neufralized as the descending wing entered deeper into the stall. Therefore, if a rapid roll occurred at low altitude and near the stall speed, recovery to level flight would be unlikely.

[ANALYSIS] The influence of ground effect on a stalled wing rolling downward, however, would probably be neufralized as the descending wing entered deeper into the stall. Therefore, if a rapid roll occurred at low altitude and near the stall speed, recovery to level flight would be unlikely. Without FDR information, the Safety Board could not determine whether the aircraft entered the stall speed range at any time. The Safety Board concluded that for roll reversals to have occurred, neither wing could have been stalled before the final roll to the right. The large roll angles reported by witnesses were probably illusionary since the accident aircraft would not have remained airborne during the large banked attitudes because of the associated loss of lift. The seratch marks on the left wingtip tank indicate that the aircraft was in a 19° left roll, a 9° noseup and a 7° right yaw attitude when the tip tank struck the runway. The small yaw angle indicates that the tip tank probably hit the runway during the first roul. Also, the relatively high pitch attitude at that time indicates that the aircraft may have "dropped out" and that the pilot may have initiated an abrupt pitch change to stop the increasing descent rate and to prevent a hard landing. According to the flight test data, stalls accelerated by landing flares can de achieved with a rapid pitch increase and result in abrupt left wing drops. A simultaneous increase of engine thrust may have accelerated the aircraft above the stall entry speed and reinstituted roll control. The changes In airfoil charactvristics and airspeed while near the stall airspeed might have made the aircraft susceptible to roll oscillations. The combination of ground effect and the increased thrust may have been sufficient to keep the aircraft above the stall speed and above the runway during the roll reversals until the bank angle and roll tate increased to the extent that the descending wing stalled at an altitude too low for recovery. 3, CONCLUSIONS Findings 1. 2. The pilots were currently certificated and qualified for the flight. There was no evidence of physical impairment or incapacitation of the pilots. The airplane was certificated and maintained according to approved procedures, There was no evidenee of a preexisting structural failure or systems malfunction. Witnesses watched the Learjet cross the runway threshold in a normal landing attitude. Seconds later, the aircraft began a series of rapid, steep roll oscillations which ended in a crash. The aircraft had flown in moderate to severe icing conditions during the descent. The flight may have encountered light icing conditions below the overcast, The pilot is believed to have used the anti-ice systems until clear of the clouds, The wing surfaces may have accumulated a Significant trace of ice after the anti-ice system was turned off, A trace of ice on the leading edge of the wings may have caused a premature stall, Wake turbulence of a departing aircraft may have initiated the rolling maneuvers of the Learjet.

[ANALYSIS] With Jimited information in hand, the Safety Board considered the possibility that the aceldent aircraft encountered the vortex or other wake turbulence from the departing Delta DC-9 and also considered the possibility that leading edge wing ice had accumulated during the descent and had affected tha flight dynamics of the landing aircraft. The Safety Board further considered the possibility that a delayed application of engine thrust during an cttempted go-round at a low speed may have induced large roll angles. Wake Turbulence At 1932:30, the Detroit local controller told the Learjet that the DC-9 would be departing on runway 9 before the Learjet landed. The DC-9's FDR speed data disclosed that the pilot of this aircraft had applied full engine thrust during a rolling takeoff at 1933743. Twenty-six seconds later when 2,200 ft down the runway, the DC-9 was rotated for takeoff. About 1934:14, the aircraft lifted off at a point 3,410 ft down the runway. Ten seconds later, about 1934:24, the landing Learjet had crossed the runway threshold while descending on a normal 3° glide slope. According to an air traffic controller, the oilot's intended touchdown point appeared to be 1,200 to 1,500 ft down the runway. The witness saw the accident aircraft cross that area at an altitude of 10 to 15 [t. Other witnesses stated that the pilot apparently initiated a go-around a few seconds later, At the time of the accident, the component of the 12~-kn wind would have moved wake vortex turbulence generated by the DC-9 at about 18 ft per second, The Learjet was obse*ved to have entered the rolling maneuver 4.5 seconds before the impact 1,640 ft past the threshold. At an average groundspeed from the threshold of 109 kns (169 ft per second), the destabilized flight maneuver probably started about 1,800 ft past the threshold. The Learjet would have reached the 1,800-ft runway position about 15 to 16 seconds after the DC-9 started rotation. The turbulence generated from the DC-9 rotation could have drifted in the wind to a point about 2,000 ft from the runway threshold. Since the precise times at which the Learjet may have been at specizic locations can be placed only within about 5 seconds, it fs conceivable that the Learjet could have encountered wake turbulence when it was first seen to roll abruptly.

[ANALYSIS] The turbulence generated from the DC-9 rotation could have drifted in the wind to a point about 2,000 ft from the runway threshold. Since the precise times at which the Learjet may have been at specizic locations can be placed only within about 5 seconds, it fs conceivable that the Learjet could have encountered wake turbulence when it was first seen to roll abruptly. The assumed filght track data of the Learjet was submitted to the Transportation Systems Center (TSC), Cambridge, Massachusetts, where it was used to determine the position relationships of the Learjet to the wake vortices generated by the DC-9, Because of the wind velocity and time separation, the TSC ruled out wake vort!zes after liftoff as a causal factor in the accident, However, the report noted that wake vortices are generated whenever an uireraft wing is creat’ ig lift and therefore wake vortices are generated between Initial rotation and uftoff. Accordingly, If vortex turbulence caused the violent rolling of the Learjet, the vortex could only have been created by the DC-9 after rotation but before liftoff. Although the strength, persistence, and flow patterns of wake 3 emir nk A RONEN RI dante aan oy, ~]9- vortices created during ground rotation are not fully known, their characteristics are being included in a current TSC data collection program. Without essential documentation, which would have been provided by an FDR, the Safety Board could not reconstruct the dynamics of the Learjet flightpath nor establish the possibility that dissipating wake vortex turbulence of the DC--9 rotation caused the éestabilization and rolling maneuvers of the Learjet, To eliminate the hazards of wake vortex turbulence, air traffic controllers are required by air traffic control handbook 7110.65A to separate arriving or departing traffic, or combinations of similar traffic, by aircraft category. In this accident, both alreraft were Category Ill aircraft and in this classification, separation should provide that the arriving aircraft does not cross the landing thresheld until the departing aircraft has crossed the runway end or a minimum distance of 6,000 ft exists between the aircraft. When the Learjet crossed the landing threshold of runway 9, the DC-9§ had not crossed the other end of the same runway; ‘owever, the distance separating the aircraft was greater than the minimum required distance, National Weather Service ground observations and upper air soundings recorded conditions associated with leing in the flight environment, About the time of the escident, several air carrier pilots reported that their aircraft had acciimulated moderate to severe airframe icing at the lower altitudes in the Detroit area, Although the cloud base was 2,300 ft above ground level and visibility was 10 mi, light icing was also reported below the overcast, Flight tests, conducted by the Safety Board, evaluated the relationships between engine power output during the descent profile and during the airspeed reduction changes of the Learjet.

[FINDINGS] The flight may have encountered light icing conditions below the overcast, The pilot is believed to have used the anti-ice systems until clear of the clouds, The wing surfaces may have accumulated a Significant trace of ice after the anti-ice system was turned off, A trace of ice on the leading edge of the wings may have caused a premature stall, Wake turbulence of a departing aircraft may have initiated the rolling maneuvers of the Learjet. During a performance Study, no abnormalities were found in the low-speed handling characteristics at speeds above the stickshaker speed, Coupled with abrupt pitch and roll control inputs during a goaround attempt at speeds within the stickshaker range, a delay fn adding thrust can result In a rapid, rolloff stall, Seeteeeeneenistnediien aeheieent ee ee ee 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of the accident was the pilot's loss of control. The loss of control may have been caused by wake turbulence of a departing aircraft, by a premature stall caused by an accumulation of wing ice, by a delayed application of eng‘ne thrust during an attempted go-around, or by any combination of these factors, 4. SAFETY RECOMMENDATIONS As a result of this accident and several others invelvirg general aviation aircraft, the National Transportation Safety Board reiterates the following recommendations made to the Federal Aviation Administration on April 13, 1978: Develop, in cooperation with industry, flight recorder standards {{FDR/CVR) for complex aircraft which are predicated upon intended aircraft usage. (Class II, Priority Action) (A-78-27) Draft specifications and fund research and development for a low-cost FDR, CVR, and composite recorder which can be used on complex general aviation aircraft. Establish guidelines for these recorders, such as maximum cost, compatible with the cost of the airplane on which they will be installed and with the use for which the airplane is intended. (Class II, Priority Action) (A-78-28) In the interim, amend 14 CFR to require that no operation (except for maintenance ferry flights) may be conducted with turbine-powered airctaft certificated to carry six passengers or more, which require two pilots by their certificate, without an operable CVR capable of retaining at least 10 minutes of intracockpit conversation when power is interrupted. Such requirements can be met with available equipment to facilitate rapid implementation of this requirement, (Class Il, Priority Action) (A-78-29) BY THE NATIONAL TRANSPORTATION SAFETY BOARD /s/ ELWOOD T. DRIVER Vice Chairman /s/ FRANCIS H. MecADAMS embder G.H. PATRICK BURSLEY Member “ \ en a Seen nee OP ees COE Te er Rs < \, statement: I fully support the report.

[ANALYSIS] The approach reference speed would have heen ebout 114 kns. However, if the pilets had used the company rule of thumb of 95 kns, based on the basic afrplane cperating weight, and added 5 kns per 1,000 lbs of fuel and 1 kn per passenger, the Vref speed for the approach would have been 109 kns. The copllot's airspeed reference bug was set at 112 kns and during the impact sequence, his airspeed indicator needle jammed at 100 kns, The captain's airspeed indicator needle was found at the zero reading and his reference "bug" was set at 101 kns, If the pilot had inadvertently monitored the malfunctioning bug for speed guidance, the error would have placed the aircraft 13 kns slower than the Aircraft Flight Manual speed and only 7 kns above the 94-kn Stickshaker actuation speed of the Stall warning system. Without the effects of ice, the speed margin errors would have been adequate for the pilot to successfully complete the landing. However, if the wing leading edges had accumulated ice in a form similar to the flight test airplane and if the stall Speed increased to 99 kns, the Learjet would have flared before landing at a speed Slightly above the ice-induced Stall speed. Under these conditions, upon flare rotation and as the wing loading and wing angle of attack were increased to lessen the descent rate, the wing inay have Witnesses described rapidly increasing engine thrust noise either before or during the rolling maneuver, One witness believed that the aircraft was accelevating as the rolling began. Postaccident engir? teardown disclosed that both engines were develcping 94 percent or greater thrust at impact. These findings indicate that, at some point in the attempted landing, the pilot decided to go-around. His decision to abandon the landing may have been due to his concern as he met dissipating wake turbulence or the decision may have been made after the left tip tank struck the runway during the first lateral upset. During the Safety Board's performance study, go-around maneuvers from 20° bank angles were simulated at altitude by simultane power, applying aft elevator control to prevent conta ¢roundplane, and applying full opposite aileron to level the airplane. On go-sround &ttempts at speeds above stickshaker activation (approximately 1.97 Vs or greater), AO problems were encountered.

[ANALYSIS] According to the captain, he reduced the throttles to flight idle well before he extended the speed brakes. Consequently, the Safety Board believes that shortly before 2147:54, the captain removed his right hand from the control wheel and used his right hand to retard the throttles to flight idle. Moreover, we believe that the aircraft was then in a substantial sideslip condit‘on which, perhaps in conjunction with some relaxation of the lateral controls or less than optimum le/t roll authority, caused the aircraft to exceed its critical controllability parameters and to roll uncontrollably into a rapid spiralling descent. As stated before, we are not ele tv fully explain hy the loss of contro: occurred. However, we note that the foregoing explanation is consiste,.t to some degree with the captain's statements about his manipulation of flight and throttle controls. Also, we believe that under the circumstances, after having apparently controlled the initial roll to the right, it would not have been unusual for the captain to have diverted his attention from the flight instruments to other instruments and controls in an effort to determine the cause of the initial roll and the cause cf the continuing airframe buffet, particularly since the other crewmembers apparently were nci aware initially of the aircraft's condition. As shown in figure 5, the second roll to a 35° right bank oncurred in about 4 seconds—a comparatively brief period in which even a slight distraction could have been critical. At the conclusion of the 4 seconds, the roll was stopped for a few seconds which indicates that lateral controls probabi,’ were applied quickly and fully in response to the comparatively rapid rate of roll. Since the captain followed the application of lateral conircls with a significant amount of rudder, as indicated by his statements, we concluce that a sideslip condition was generated which placed the aircraft in a laterally uncontvo. lable condition as evidenced by the resumption of the roll to the right. Further, it is possible that cocking or deflection of the No. 7 slat added to rolling moment imbalance at this critical point. Thereafter, the speeds and angles of attack generated by the rapid descer* and high g-forees combined with the extenJed No, 7 slat to keep the aircraft in an uncontrollable condition until the slat was torn from the wing. During the investigation, questions were raised about why the flightcrew might have extended the leading edge slats under the existing operating conditions.

[ANALYSIS] The No. 7 sict failed to retract. About 2147:45, the aircraft reached abcut 35° of right bank where the captain disconnected the autopilot and rapidly volled the airceaft to the left to a near wings-icvel attitude. Tne aircraft could have been stabilized in wing-level flight witi: appropriate deflection of the lateral controls. About 2147:47, the aircraft again beown te roll to the right, probably while the captain was distcz ted by activities related to the isolation of the No. 7 slat in the extended position. Shortly before 2147:51, the captain recognized the rapid right roll, and he rapidly applied full deflection of the lateral controls to stop the oll. Tiie roll was stopped near 35° of right bank for several seconds during which the captain removed his right hand from the control wheel, pulled the throttles to flight idle, and deflected full or nearly full left rudder to augment lateral controls. In response to the rapid and full or nevrly full deflection of the flight controls, the aircraft entered a substantial right sideslip. The sideslip combined with the aircraft's mach number and angle of attack tu reduce the lateral control margin to zero or less. The airciuft resumed the right roll and began to descend rapidly and uncontrollably. The captain extended speed brakes, detected no reaction, and retracted them. About 2148:25, the aircraft completed {{60° of roll while descending to about 21,000 feet. Shortly thereafter, the captain commanded landing gear extension which was accomplished by the first officer. The aircraft continued to descend rapidly, and it continue’ to roll to the right until the No. 7 slat was torn from the wing and): teral control was restored. About 2148:58, the captain regained cc trol of the aircraft at an altitude of about 8,000 feet. Since our weighing of the evidence involves a rejection of the possibility of an unscheduled extension of the No. 7 slat and a partial rejection of the captain's recollection of his actions following extension of the slats, the Safety Board believes that the following comments are appropriate: We believe the captain's erasure of the CVR is a factor we cannot ignore and cannot sanction. Although we recognize that habits can cause actions nct desired or intended by the actor, we have ‘difficulty accepting the fact that the captain's putative habit of routinely erasing the CVR after each flight was not restrainable after » flight in which disaster was only narrowly averted. Our skepticism persists even though the CVR would not have contained any contemporaneous information about the events that immediately preceded the loss of control because we believe it probable that the 25 minutes or more of recording which preceded the landing at Detroit could have provided clues about causal factors and might have served to refresh the flighterew's memories about the whole matter. 3. CONCLUSIONS

[ANALYSIS] When mach number or angle of attack was below these values, the aircraft should have been controllable laterally and when exceeded, the aircraft would have been uncontrollable laterally. The reason for the lack of controllability when the critical values are exceeded is the severe disruption of airflow and associated loss of lift that occurs over the wing area aft of the extended slat. The loss of lift creates a rolling moment that exceeds the countermoment the oilot can produce with full deflection of lateral controls. The aircraft will respond to the imbalance of rolling moments by rolling uncontrollably toward the wing with the extended slat. By omitting any effecis of sideslip, the calculated values of mach number and anyle of attack estabiished that the accident. aircraft was laterally uncontrollable after 2148:07. At that time, the aircraft's indicated airspeed was about 270 knots (0.83 mach), its altitude was about 36,500 feet, and its rate of descent was in excess of 34,000 feet per minute. After the No. 7 slat was torn ‘rom the aircraft, lateral control was restored and the captain was able to roll the wings parallel to the horizontal and recover from the spiral dive. Simulations of the spiral dive confirmed that loss of the slat restored lateral control and made recovery possible. Also, simulations indicated that although extension of the landing geer significantly reduced the aircraft's speed (and mach number), recovery would have been doubtful without loss of the slat because of the high angles of attack which were developed during the latter part of the descent. Because of the conflict between the captain's assertions regarding the ineffectiveness of lateral controls and the aerodynamic evidence related to aircraft performance and controllability, significant efforts were made to determine the aircraft's actual motions and performance”, as reflected in its FDR traces, subsequent to the first observed anomalies in the g-trace. These efforts included additional heading gyro tests, extensive calculaticns of gimbal error associated with heading gyro performance, and flight tests, Analysis of flight test and flight simulator data indicated that high rates of heading change, such as the 5° in 0.5 second that oceurred about 2147:45 in the accident aireraft's heading trace, could not be achieved unless the aircraft was in a banked attitude with high load factors applied. Since the load factor was about 1.0 g at thai time, the rapid heading change could not have been associated with turning flight. Also, during flight tests, a full rudder deflection sideslip, when released, did not produce such a rapid heading change, and during sudden extensions of the No. 7 slat in flight simulations, the heading changed comparatively slowly to the right. However, the heading gyro tests and heading gyro gimbal error calculations showed that such an apparent rate of heading change eould have beer produced by a rapid roll to the left. Consequently, it was hypothesize? that during the preceding 12-second period between 2147:33 and 2147:45, the aircraft had rolled to the right.

[FINDINGS] When scheduled to re-ract, the No. 7 leading edge slat failed to retract probably because tensile forces created by air loads combined with friction and side forces on the piston rod, caused by preexisting miscasignmen* of the slat, exceeded the available hydraulic retraction force. The No. 7 leading edge slat in the extended position created rolling inoments to the right that could have been countered with about 46° of control wheel deflection to the left; an additiono, 13° of control wheel Gefiection would have been needed to counter moments associated with a l-ineh upware float of che right outboard aileron. After recognizing the right roll condition, the captain rolled the aircraft to a near wings-level upright positien; thereafter, through untimely use of the flight controls, he permitted the aircraft to roll to the right into an uncontrollabie attitude. The captain probably was distracted immediately after restoring the aircraft to near level flight by his efforts in attempting to rectify the source of the control problem. The captain probably induced sideslip shortly before 2147:54 when the aircrafi was at mazh 0.79, an angle of attack of 5.7°, and an angle of bank of about 35° to the right. A sideslip angle of 4.6° to the right could have caused the aircraft to become laterally uncontrollable. The aircraft descended in a spiral dive from 39,000 to about 5,000 feet in 63 seconds; during the descent, the aircraft's speed increased to a maximum speed of about 0.96 mach at 31,300 feet. When the aircraft's speed exceeded 0.83 mach and its angle of attack exceeded 6° near 36,500 feet, the rolling moments caused by the extended No. 7 slat suostantially exceeded the maximum available lateral control authority at 0° of sideslip. The aireraft was not ecntrollable during its descent below «"sout 36,500 feet un:il the No. 7 leading edge slat separated from the right wing. Vertical acceleration forces increased throughout the spiral descent to a maximum: of about 6.0 g's during the recovery. The accident was survivable. 22. Minor injuries to passengers were caused by the g-forces. 3.2 Probable Cause The Safety Board determines that the probable ceuse of this accident was the isolation of the No. 7 leading edge slat in the fully or partially extended position after an extension of the Nos. 2, 3, 6, and 7 leading edge slats ard the subsequent retraction of the Nos. 2, 3, and 6 slats, and the captain's untimely flight control inputs to counter the roll resulting from the slat asymmetry.

[ANALYSIS] If a 5° right sideslip is introduced, the local critical mach number is reached at a freestream mach number of 0.787. Consequently, with the introduction of sideslip, the accident aircraft could have reached critical controllability parameters at freestream mach numbers significantly below the critical values for 0° of sideslip. ~31- Flight test data and flight simulations indicated that in the B-727 full deflection of lateral controls unc full deflection of rudder in the same direction can produce sideslip angles cf 4.5° to 6.5°. Also, flight tests in conditions similar to Flight $41's at 39,000 feet showed that rolls with full deflection of the lateral controls produced sideslip angles of about 5° in the direction opposite to the roll, As shown in figure 5, about 2147:47, after a left roll to a near wings-level position, the aircraft again began to roll to the right and about 4 seconds later was at 35° of right vank where the roll was checked for a few seconds. Under the cir:umstances, we believe this roi probably occurred while the captain was distracted by activities related to the No. 7 slat having been isclated in the extended or partially extended position. Thereafter, if full laterat and rudder controls were applied simultaneously cr in rapid succe<sion to stop the roll, significant sideslip could have been introduced at a critical time, and the aireraft could have become laterally uncontrollable well before its 0° sideslir controllability parameters were reached, Calculations of mach numoer and angie of attact., which take into account 4 recorder error of -7.5 knots and the effects of roll rate on angle of attack, indicate that at 2147:51, the accident aircraft was at mach 0.79 and an angle of attack of §.7°. Also, according to ginbal error calculations, the aircraft was banked at out 35° to the right. Under these conditions, if a 4.8° right sideslip angle was introduced, lateral control could have been lost. About 2147:54, the aircraft's rate of descent began to increase very rapidly, although FDR indicated airspeed was stable at 236 knots, which indicates that. thrust was subs’antially reduced or drag was substantially increased shortly before that time. According to the captain, he reduced the throttles to flight idle well before he extended the speed brakes. Consequently, the Safety Board believes that shortly before 2147:54, the captain removed his right hand from the control wheel and used his right hand to retard the throttles to flight idle.

[ANALYSIS] According to flight tests, buffet begins and is shown in the g-trace about 2 seconds after the beginning of leading edge slat extension. Therefore, it appears that slat extension began about 2147:32. Assuming © reaction time of 2 seconds to retract the Slats, retraction would have begun about 2147:39 and would have been completed 5 to 6 seconds later. This would mean that slat asymmetry would have begun about 2147:40 _ which would have caused an increasing rate of roll to the right while the Nos, 2, 3, and 6 Slats retracted. Gimbal error calculations indicate that from 2147:40 to 2147:45 the aircraft rolled from about 10° of right bank to about 35° of right bank at an increasing roll rate. This is compatible with the increasing slat asymmetry that would have occurred during that 5 seconds. Given the foregoing assessment of the aircraft's motions and performance, it is apparent that the aircraft was initially controllable following isolation of the No. 7 leading edge slat in the extended position because, beginning about 2147:45, the aircraft was rolled to the left to a near wings-level position. About 4 seconds later, the aircraft was again banked to the right about 35°. However, following a 2-second pause at 35°, the eireraft resumed its roll to the right and it began to descend rapidly. The Safety Board is not able to determine conclusively why the captain failed to retain control of the aircraft after having once rolled it from a 35° right bank to near a wings-level attitude. Although we cannot positively exclude spatial disorientation of the captain as a possible reason for his failure to retain control, we believe it more probable that a number of factors combined to place the aircraft in an attitude where critical controllabijity parameters were exceeded well before the parameters established from subsequent flight simulations of the maneuver. These factors involve the actual effectiveness of the lateral controls, the actual margins of lateral control with an extended No, 7 slat, a cruise mach number tiat we believe was higher than 0.80, the effects of sideslip induced by full deflection of lateral controls and rudder, and distraction of the captain. It is possible, for instance, that because of aileron and spoiler rigging tolerances, the accident aircraft had tess than specified left roll capability. Although these tolerances would not have been noticeable in normal maneuvering flight, they could have become a factor during full deflection of the controls. Also, it is possibie that lateral control margins were reduced to values similar to those shown in figure 4, rather than the higher values used in the flight simulaticns. The Safety Board believes that Flight 841's cruise mach number at 39,000 feet pressure altitude probably was higher than 0.80. Support for a higher cruise mach number is indicated by a comparison of the airspeed and altitude traces with the captain's statements about mach number and airspeed.

[ANALYSIS] However, the heading gyro tests and heading gyro gimbal error calculations showed that such an apparent rate of heading change eould have beer produced by a rapid roll to the left. Consequently, it was hypothesize? that during the preceding 12-second period between 2147:33 and 2147:45, the aircraft had rolled to the right. Based on heading gyro tests and actual heading, bank, and pitch angles vecerded during flight simulations of Flight 841's maneuver, it was determined that indicated heading could be accurately calculated by using the standard mathematical equation for heading gyro errors. By use of pitch angles and actual heading angles from the simulations, calculations of gimbal error associated with various bank angies produced an indicated heading trace which was comparable to the accident aircraft's FDR trace with the time shift removed. Calculations of vank angles up to 60° were further confirmed by using aircraft turning performance equations and FDR values of normal acceleration, altitude, and airspeed. These data were programed into the Safety Board's scientific data reducticn and plotting computer which drew the results for the first 360° of roil shown in figure 5. As shown in figure 5, the calculated indicated heading trace for these heading, pitch, and bank angles compares almost exactly with the accident aircraft's FOR heading trace with the time shift removed. Beginning about 2147:34, the aircraft oegan to roll slowly to the right, and about 6 seconds later, the rate of roll began to increase. At 2147:45, the aircraft was in a right bank of about 35°, after which it was rolled rapidly to the left to near a wings-level attitude. About 2147:47, the aircraft began to roll again to the right, and the roll was arrested briefly at 2147:51 near 35° of right bank. At 2147:53, the aircraft resumed its roll to the right and about 2148:07, the aircraft was inverted with @ pitch attitude about 45° below the horizontal. As shown in figure 5, the aircraft's actual heading (calculated) changed comparatively little during the 360° of roJl. Based on flight test data and known times for extension and retracticn of leading edge slats, the 12-second period between 2147:33 and 2147:45 was examined to determine whether a compatible relationship existed among slat extensicn/retraction cycles, slat asymmetry, and calculated angles of baiik that were achieved during the period. According to flight tests, buffet begins and is shown in the g-trace about 2 seconds after the beginning of leading edge slat extension. Therefore, it appears that slat extension began about 2147:32. Assuming © reaction time of 2 seconds to retract the Slats, retraction would have begun about 2147:39 and would have been completed 5 to 6 seconds later.

[FINDINGS] Our skepticism persists even though the CVR would not have contained any contemporaneous information about the events that immediately preceded the loss of control because we believe it probable that the 25 minutes or more of recording which preceded the landing at Detroit could have provided clues about causal factors and might have served to refresh the flighterew's memories about the whole matter. 3. CONCLUSIONS Findings 1. The flighterew was properly certificated and was qualified for the flight; the captain had requalified in the B-727 only recently, had flown 21 hours 50 minutes since requalifying, had not flown for about 3 months before requalification, and had flown exclusively as a first officer on B747's from November 1977 to December 1978. The aircraft was certificated and maintained in accordance with existing regulations and procedures, The inboard slat track T-bolt on the No. 7 leading edge slat had failed predominantly in fatigue. the right outboard aileron actuator hinge fitting bolt rib had failed predominantly in fatigue. The wear pattern on the slat alignment hooks indicated that the No. 7 leading edge slat was not aligned properly. There was no other evidence of irregularity, malfunction, or failure of the aircraft's flight control, autopilot, hydraulic, or flep systems that might have caused or contributed to 4 lateral control problem. The aircrait's gross weight and center of gravity were within the authorized performance and maneuvering envelopes when a lateral control problem develuped. The aircraft was cruising at 0.816 mach in level flight and smooth air at 39,000 feet when a lateral control problem developed. A failure of the right outboard aileron actuator hinge fitting bolt before development of the lateral control problem would have permitted the aileron to float ucward about 1 inch; this condition would have required about 13° of left \leflection of the control wheel to maintain wings-level flight and would have been noticeable. A right roli and a lateral control problem were caused by isolation of the No. 7 leading edge slat in the extended position. There was no evidence of any combination of failures or malfunctions in the aircraft's flight control system that would have caused an unscheduled extension of the No. 7 leading edge slat. The Neos, 2, 3, 6, and 7 leading edge slats were scheduled to the extended position, and the Nos. 2, 3, and 6 slats were retracted as a consequence of the flightcrew's actions. When scheduled to re-ract, the No. 7 leading edge slat failed to retract probably because tensile forces created by air loads combined with friction and side forces on the piston rod, caused by preexisting miscasignmen* of the slat, exceeded the available hydraulic retraction force.

[PROBABLE CAUSE] However, I am troubled by the fact that the Board i:as categorically rejected the crew's sworn testimony without the crew having had the opportunity to be confronted with all of the evidence voon which the Board was basing its findings. Ai the time of the first deposition, the following evidence was not available to the crew or to the Board: the flight data recorder analysis, the results of the simulator and flight tests, and the tilt table tests. Although the crew was deposed a seco,:3 time, their testimony was limited to one issue, i.e., the physical location of the flight engineer at the time of the incident. ! had recommended that since the Board was ordering a second deposition it be conducted de novo so that the erew would have been aware of all the evidence. The Buard did not agree. Furthermore, I do not agree that a prooable cause of this accident, as stated by the Board, was “the captain's untimely flight control inputs to counter the roll resultin from the slat asyminetry.” In my opinion, the <aptain acied expeditiously and reasonably in attempting to correct for the severe right roll condition induced by the extended slat. JAMES B. KING, Chairman, and PATRICIA A. GOLDMAN, Member, did not participate. June 9, 1981. 5. APPENDIXES APPENDIX A INVESTIGATION AND HEARING 1. Investigation The Safety Board was notified of the accident about 0200 on April 5, 1979. Aninvestigator from the Chicago, Iinois, Field Office was sent immediately to Detroit, Michigan; operations, systems, and sti uctures investigators were sent from the Headquerters office. Later, investigative responsibility for the accident was transferred to the Safety Board's Headquarters in Washington, D.C. Representat. ves from the Federal Aviation Administration, Trans World A. lines, Ine., Boeing Company, and the Air Line Pilots Association participated in th: investigation. 2. Public Hearing There was no public hearing. The fiighterew was deposed in Los Angeles, California, on April 12, 1979. Two FAA inspectors, three flight attendants, and the flighterew were deposed in Kansas City, Missouri, on January 29, 1980. -3a8- APPENDIX B PERSONNEL INFORMATION Captain Harvey G. Gibson Captain Gibson, 44, was employed by Trans Wer‘, Airlines, Inc., on December 9, 1963, He holds Airline Transport Pilot Certificate '.o. 1192040 with an airplane multiengine iand rating and type ratings in the DC-9, B-727, B-747, 1-1011, and commercial pilot privileges for airplane single engine land, rotorcraft-helicopter, and balloons. His first-class medical certificate was issued March 7, 1979, with no limitations.

[ANALYSIS] NTSB Aviation Accident Report 45 those that the accident pilots experienced on the night of the accident are extremely rare, an individual pilot encountering such strong and gusty winds would have little or no past experience on which to draw during the encounter. The NTSB concludes that the unexpectedly strong and gusty crosswinds the airplane encountered as it accelerated during the takeoff roll made maintaining directional control during this takeoff a more difficult control task than the captain was accustomed to dealing with; however, had the captain immediately reapplied significant right rudder pedal input as the airplane was continuing its left turning motion, the airplane would not have departed the runway. 2.2.2 After the Airplane Left the Runway The airplane departed the left side of runway 34R at 1818:17, and, based on postaccident interviews and FDR data, the pilots began to experience sharp vertical accelerations as the airplane traversed the uneven terrain beyond the edge of the runway. Although it could not be determined how long the captain continued to manipulate the nosewheel steering tiller (tiller position is not recorded by the FDR), FDR data did show that the captain maintained full right control wheel inputs for about 1 second after the airplane left the runway. At 1818:20, the captain began to reduce power, and, about 1 second later, he called for a rejected takeoff. The pilots began to apply the brakes and moved the thrust levers to produce reverse thrust. Investigators noted that about 5.75 seconds passed from the time it became apparent to the captain that the airplane was going to leave the runway (at 1818:14.25, less than 1 second after the “Jesus” comment and after the captain started making ineffective control wheel and nosewheel steering tiller inputs) to the captain’s initiation of a rejected takeoff (at 1818:20). A study published by the British Air Accidents Investigation Branch that examined the time a pilot requires to begin a rejected takeoff after a sudden unexpected event during the takeoff roll found average response time to be 2.7 seconds, with a range of 1.5 to 4 seconds.87 2.3 Wind Information and Runway Selection Based on this research, the captain’s initiation of the rejected takeoff procedure occurred about 3 seconds later than average response and about 1.75 seconds beyond the top of the range. The captain’s actions during his final attempts to keep the airplane on the runway were ineffective, but they also delayed him from performing a rejected takeoff. Because of the positioning of the nosewheel steering tiller and the control wheel and the captain’s simultaneous use of these two controls, the captain did not have his right hand on the thrust levers during the excursion. The need to reposition his right hand to the thrust levers as the airplane was bouncing along uneven ground also likely contributed to the delay.

[ANALYSIS] Furthermore, because of slight delays in the effect each rudder pedal adjustment had on the airplane’s rate of heading change, the captain had to anticipate the effect of each adjustment ahead of time. This task was very difficult for the captain because of the highly variable and unpredictable nature of the crosswind gusts. The first of the captain’s large right rudder pedal inputs resulted in an apparent overcorrection of the airplane’s heading after about 1.5 seconds. Because the captain had no way to distinguish the effect of his inputs from the effect of the variable crosswind component, he likely believed that this large right rudder pedal input exceeded the amount of rudder correction that would be required to compensate for the crosswind. (In fact, the NTSB’s airplane performance analysis revealed that the slight overshoot resulted from a decrease in the crosswind component of about 11 knots.) 85 Although investigators were unable to attribute the comment definitively to either pilot, the sound of the pilot’s voice indicates that it was most likely the captain. He would have been acutely aware of the relationship between his pedal inputs and the airplane’s heading oscillations, whereas the first officer was primarily monitoring the airplane’s airspeed indicator at the time. NTSB Aviation Accident Report 44 abortive right rudder pedal input at 1818:14, the captain stopped making rudder pedal inputs entirely. Simultaneously, he began to apply right control wheel (aileron) inputs (instead of maintaining left control wheel inputs, which would have been appropriate for the left crosswind conditions), and he tried to steer the airplane to the right using the nosewheel steering tiller.86 Although the FDR does not directly record tiller inputs, FDR data, airplane performance analysis, and physical evidence indicate that the captain did, in fact, turn the control wheel and the steering tiller to the right about 2.75 seconds before the airplane ran off the runway, and neither input had a significant effect on the airplane’s heading. (Figure 1, in section 1.1 of this report, shows the oscillations in the airplane’s heading relative to the baseline heading and the percentage of forward displacement of the right rudder pedal between 1818:01 and 1818:15, ending about 2 seconds before the airplane left the runway.) The captain’s use of the nosewheel steering tiller was contrary to company procedures and neither of these late control inputs was an effective method for turning the airplane at high speed. The NTSB concludes that the captain’s use of tiller and full right control wheel in the 3 seconds before the excursion likely resulted from acute stress stemming from a sudden, unexpected threat, perceived lack of control, and extreme time pressure. The comment “Jesus” suggests that, by 1818:13.5, the captain was aware that his second large rudder pedal input and its subsequent relaxation were not reversing the direction of the airplane’s turn as the first large rudder input cycle had. However, the captain did not immediately apply another large right rudder pedal input; rather, he relaxed the rudder pedal to its neutral position.

[ANALYSIS] The comment “Jesus” suggests that, by 1818:13.5, the captain was aware that his second large rudder pedal input and its subsequent relaxation were not reversing the direction of the airplane’s turn as the first large rudder input cycle had. However, the captain did not immediately apply another large right rudder pedal input; rather, he relaxed the rudder pedal to its neutral position. Findings from the NTSB’s operational study of simulated takeoffs in a steady 35-knot direct crosswind in a 737-500 simulator suggested that, if the captain had renewed vigorous application of right rudder pedal by 1818:14, he would likely have been able to counteract the airplane’s left turn and keep the airplane on the runway. (Although it was not possible to replicate the gusty crosswind conditions that affected the accident airplane, the success of the captain’s first large rudder pedal input and relaxation thereof indicates that he might have been able to counteract the airplane’s left-turning motion.) However, the NTSB’s operational study also showed that it was probably too late for him to prevent an excursion by 1818:15. (The airplane ran off the left side of the runway about 1818:17.) Several ATP-rated pilots who participated in postaccident flight and engineering simulator studies of the accident takeoff (four of whom had 737 type ratings) concurred with this analysis. Continental 737 operating manuals state that the nosewheel steering tiller should be used to turn the airplane only at slow taxi speeds (up to about 20 knots) and that rudder pedal steering and inputs to the rudder surface are more effective means of keeping the airplane on the runway centerline during the takeoff roll. The captain told investigators that he used the airplane’s steering tiller (to no avail) in a brief, desperate attempt to keep the airplane on the runway. Postaccident review of Continental’s operational flight data for crosswind takeoffs that was collected during about 8 years of operation indicated that it is rare for Continental’s pilots to encounter crosswind components of 30 knots or more. The data showed only 62 such occurrences (4 of which were in 737-500 airplanes) in 940,000 takeoffs. Takeoffs in gusty winds were not specifically identified. Because operational flight data indicate that winds as strong as 86 During postaccident interviews, the captain explained that he input right aileron because he was concerned about keeping the airplane upright in the terrain on the left side of the runway and that he reached for the nosewheel steering tiller in a final effort to keep the airplane on the runway. NTSB Aviation Accident Report 45 those that the accident pilots experienced on the night of the accident are extremely rare, an individual pilot encountering such strong and gusty winds would have little or no past experience on which to draw during the encounter.

[CONCLUSIONS > FINDINGS] The unexpectedly strong and gusty crosswinds the airplane encountered as it accelerated during the takeoff roll made maintaining directional control during this takeoff a more difficult control task than the captain was accustomed to dealing with; however, had the captain immediately reapplied significant right rudder pedal input as the airplane was continuing its left turning motion, the airplane would not have departed the runway. NTSB Aviation Accident Report 60 11. The captain’s initiation of a rejected takeoff was delayed by about 2 to 4 seconds because he was occupied with the nosewheel steering tiller and right control wheel input, both of which were ineffective and inappropriate for steering the airplane. 12. If air traffic control personnel and pilots operating at airports located downwind of mountainous terrain had sufficient airport-specific information regarding the localized and transient nature of strong and gusty winds associated with mountain wave and downslope conditions, they would be able to make more informed runway selection decisions. 13. Although the Denver International Airport air traffic control tower local controller followed established practices when he provided the accident pilots with the runway 34R departure end wind information with their takeoff clearance, he did not (nor was he clearly required to) provide information about the most adverse crosswind conditions that were displayed on his ribbon display terminal; therefore, the pilots were not aware of the high winds that they would encounter during the takeoff roll. 14. If the Federal Aviation Administration had published the required letter to airmen describing the sensor locations, operational capabilities, and limitations of the low-level windshear alert system (LLWAS) at Denver International Airport and the accident pilots had been familiar with its content, they might have been more likely to request additional LLWAS sensor wind information when they saw the clouds moving swiftly across their departure path before they accepted their takeoff clearance and/or began their takeoff roll. 15. Although the departure wind information the captain received with the takeoff clearance from the Denver International Airport (DEN) air traffic control tower (ATCT) local controller indicated that the winds were out of 270° at 27 knots (which resulted in a stronger-than-expected 26.6-knot crosswind component), the reported winds did not exceed Continental’s maximum crosswind guidance of 33 knots, and the captain could reasonably conclude that the winds, as reported by DEN ATCT, did not exceed either his or the airplane’s crosswind capabilities. 16.

[ANALYSIS] As the airplane continued off the runway, the captain initiated a rejected takeoff; the airplane subsequently came to rest in a field between runways 34R and 34L. The remainder of this section evaluates the pilots’ decisions and actions during the attempted takeoff and after the airplane left the runway. NTSB Aviation Accident Report 42 2.2.1 Attempted Takeoff The airplane was nearly aligned with the runway centerline when the captain began to advance the thrust levers for takeoff (at 1817:45). The captain stated that, as the airplane accelerated, he shifted the primary focus of his attention from the thrust setting to outside visual references and kept the airplane tracking along the runway centerline. (As the airplane accelerated through about 50 knots, a baseline heading of about 346° resulted in the airplane tracking along the runway centerline.) Meanwhile, according to postaccident interviews, the first officer’s attention was primarily focused on monitoring the engine instruments, consistent with company policy. At 1818:04, when the airplane’s speed was about 57 knots, the CVR recorded the first officer advising the captain that the engines had stabilized at takeoff thrust. The first officer stated that, after this point, he shifted his attention to monitoring the airspeed so that he could make the standard airspeed callouts, the first of which was at 100 knots. FDR heading and rudder pedal position data showed that, as the airplane accelerated between 1818:01 and 1818:07, the captain gradually increased the right rudder pedal input to almost 50 percent of its forward range of travel. The captain’s advancement of the right rudder pedal during this time was not perfectly smooth, likely because, as the aerodynamic effectiveness of the rudder increased, he was making incremental adjustments in an attempt to find the correct amount of right rudder pedal input to compensate for the variable crosswind as the rudder became effective. The airplane heading moved left by a fraction of a degree at 1818:03 and again at 1818:05, but it returned to the baseline heading each time as the captain continued to advance the right pedal. At 1818:06.7, the captain applied a large and rapid right rudder pedal input (reaching about 88 percent of the pedal’s available forward range)84 The captain likely anticipated the heading change back to the left because he had begun to advance the right rudder pedal again just before this change in the direction of movement occurred. He paused his right rudder pedal input briefly at 53 percent. Then, when the airplane’s nose crossed the baseline heading from right to left without slowing, he further advanced the right rudder pedal to 72 percent, reaching this position at 1818:11.75. He began to relax this right rudder pedal input immediately (as he had after reaching the 88-percent input); about that time, the main landing gear tires began to skid on the runway pavement.

[ANALYSIS] Then, when the airplane’s nose crossed the baseline heading from right to left without slowing, he further advanced the right rudder pedal to 72 percent, reaching this position at 1818:11.75. He began to relax this right rudder pedal input immediately (as he had after reaching the 88-percent input); about that time, the main landing gear tires began to skid on the runway pavement. The leftward change in heading almost stopped at 1818:12.2, but the airplane began to turn rapidly to the left again at 1818:13.2. The captain had been steadily relaxing the right rudder pedal, reaching 36 percent at this time; however, as the airplane began turning rapidly left this time, he abruptly relaxed the and then promptly began to relax the right rudder pedal input. During this input, the airplane’s heading moved a fraction of a degree left of the baseline heading due to the crosswind and then began to move back to the right (about 1/2 second after the peak of the input). The captain continued to relax the right rudder pedal input as the nose of the airplane moved to the right. After the airplane’s heading crossed to the right of the runway baseline heading, the captain relaxed the right rudder pedal input further to about 15 percent. Shortly thereafter, the airplane’s heading reached about 1° right of the baseline heading and then (at 1818:10.2) began to move back to the left. 84 Simultaneous with this rudder application, the captain added left control wheel (aileron) inputs (consistent with a normal crosswind takeoff technique) and a slight aft control column input. NTSB Aviation Accident Report 43 right pedal to its neutral position. Simultaneous with this sudden relaxation (at 1818:13.5), one of the pilots (likely the captain) stated, “Jesus.”85 The airplane’s response to his earlier large rudder pedal input likely led the captain to expect that a slightly smaller right rudder pedal input would stop a subsequent left turn and that the correction would again be evident after a slight delay. As a result, the next time the nose of the airplane began to drift left, the captain used a smaller, but still substantial, right rudder pedal input and again relaxed that rudder pedal input in anticipation of a delayed effect. However, this time, the airplane did not respond as he probably expected. Instead of reversing direction after a slight delay, the airplane’s left turn only slowed and then rapidly resumed, which led the captain to believe that the airplane was not responding to his rudder input.

[ANALYSIS] NTSB Aviation Accident Report 52 Although during postaccident activities investigators described attempted simulator takeoffs in direct 35-knot crosswinds as only slightly difficult,97 Increased training in this area could benefit pilots because it could help them identify how wind characteristics may affect airplane response and how pilot technique may affect steering difficulty. However, limitations in existing simulator capabilities are an obstacle to providing pilots with realistic gusty crosswind training. Although much work has been done to improve the fidelity of flight simulators in recent decades, the NTSB is unaware of any recent efforts to improve the fidelity of the wind models used in simulators for the training of gusty crosswind takeoffs and landings. these assessments did not adequately reflect most real-world, high-crosswind takeoffs because Continental’s 737-500 simulators do not incorporate wind gusts. Further, takeoff data obtained from Continental indicated that the company’s pilots rarely, if ever, encountered crosswind components greater than 30 knots during actual flight operations. It is unlikely that Continental’s pilots were proficient at handling strong and gusty crosswinds like those encountered by the accident pilots during their takeoff roll. Steering control dynamics are quite different when taking off in steady wind conditions as compared to gusty crosswind conditions. A takeoff in a steady crosswind requires a pilot to compensate for gradual changes in the airplane’s tendency to turn into the wind by testing to see how much rudder correction is needed and slowly adjusting to match slow changes in the required amount of rudder correction. The required amount of rudder correction changes relatively slowly and follows a predictable pattern. According to a Boeing study of 737-500 crosswind takeoff performance, the amount of rudder pedal input needed to keep the airplane tracking the runway centerline during a steady crosswind takeoff varies as a function of airspeed and crosswind component, with the amount of rudder correction needed increasing up to a certain airspeed and diminishing gradually thereafter. Although a pilot may identify the proper rudder position by moving the rudder pedals back and forth, or “bracketing” the target position, and observing the effect on the airplane’s tracking of the runway centerline, the required amount of rudder correction changes slowly and predictably, so the task is not very difficult. By contrast, during takeoffs in strong and gusty crosswinds, a pilot must do all of the above while simultaneously compensating for disturbances in heading caused by fluctuations in the magnitude of the crosswind component. In these conditions, it can be more difficult to determine whether a deviation in the airplane’s heading is the result of a change in the crosswind component or the slight lag in the effect of a prior rudder input. Airplane control dynamics may also be affected by the magnitude or frequency of pilot control inputs. Although some bracketing of the target rudder position is necessary in both steady and gusty crosswind conditions, bracketing with control inputs that oscillate too much or too slowly when taking off in very strong and gusty wind conditions may increase the risk of pilot confusion about the relationship between control inputs and airplane response.

[ANALYSIS] However, this time, the airplane did not respond as he probably expected. Instead of reversing direction after a slight delay, the airplane’s left turn only slowed and then rapidly resumed, which led the captain to believe that the airplane was not responding to his rudder input. The NTSB’s aircraft performance study indicated that, as the captain began his second large right rudder pedal input, the left crosswind component increased from 30 knots to about 40 knots and that, as he began to relax this rudder pedal input, the crosswind component increased above 40 knots for about 1.5 seconds. (The study showed a peak wind of about 45 knots at 1818:12.) This strong crosswind gust increased the airplane’s sideslip angle, significantly reduced the effect of the captain’s 72-percent rudder pedal input, and weathervaned the airplane’s nose further to the left. The airplane’s heading continued to move left of the baseline heading at an increasing rate; however, instead of making another significant right rudder pedal input (the only control input that could have corrected the airplane’s leftward veer off the runway), the captain began to make disorganized and ineffective control inputs. For example, while the airplane’s nose was moving rapidly to the left, increasing right rudder would be the most effective control strategy, yet the captain completely relaxed the right rudder pedal input. FDR data showed that, after one small, The two unusually large (88 and 72 percent) right rudder pedal inputs made by the captain between 1818:06.7 and 1818:13.2 were of similar duration (about 3 seconds for each complete cycle of rudder input and relaxation). As shown in figure 9, these oscillatory right rudder pedal inputs were similar to each other in shape and differed from the smaller, incremental adjustments the captain made earlier in the takeoff roll. The captain’s switch to unusually large inputs changed the dynamics of the situation in ways that may have made it more challenging for him to subsequently control the airplane’s heading and track the runway centerline. For example, to avoid overshooting the baseline heading after each large right rudder pedal input, the captain had to compensate by relaxing the right rudder pedal more than he would have had to for a smaller rudder pedal advancement. Furthermore, because of slight delays in the effect each rudder pedal adjustment had on the airplane’s rate of heading change, the captain had to anticipate the effect of each adjustment ahead of time. This task was very difficult for the captain because of the highly variable and unpredictable nature of the crosswind gusts. The first of the captain’s large right rudder pedal inputs resulted in an apparent overcorrection of the airplane’s heading after about 1.5 seconds.

[ANALYSIS] The captain agreed with the first officer, but it is unclear from the exchange whether the captain recognized that she had previously misspoken. From the beginning, when the flight crew received the taxi clearance, there were instances in which the captain and the first officer had disparate thoughts about the clearance yet did not make effective use of available resources to verify the information. In one instance, the captain indicated that she believed that they were instructed to hold short of the active runway (of which she misspoke and stated was runway 22 instead of 23), but the first officer indicated, “I think he said … that we could cross it.” In this instance, the first officer was correct, and the captain went along with what the first officer stated. During a subsequent discussion (in which the captain and the first officer had different ideas about which direction to taxi), the captain again went along with what the first officer thought to be accurate; however, in this case, the first officer was incorrect, which resulted in the captain making a wrong turn onto the taxiway. In either case, because the crewmembers were not in agreement about the controller’s instructions, they should have verified the correct clearance with the controller. In addition, there were several instances during the taxi in which the captain or the first officer read back incorrect information, and the mistake went uncorrected by the other crewmember. Just before initiating the takeoff roll, the captain asked the first officer to request a wind check. However, the captain repeated the winds back to the first officer incorrectly, and the first officer agreed; thus, neither the captain nor the first officer was effectively listening to each other or the controller. The captain’s next statement that the winds were “pretty much straight down” the runway was also incorrect, and the first officer did not correct that error. Although none of the incorrect information during the taxi related directly to the circumstances of the accident, the exchanges are of concern in that, collectively, they provide evidence that neither crewmember was particularly focused. The CRM skills exhibited by the flight crew were inconsistent with the skills needed for effective communication and the coordination of a professional flight crew. During the taxi, the captain’s casual tone and lack of leadership and the flight crew’s inattention to details foreshadowed elements of the crew’s subsequent performance in responding to the anomaly. The captain’s lack of accuracy in her pretakeoff briefing and the first officer’s indirect questioning of it were missed opportunities for the crew to make use of the briefing for its intended purpose as a CRM tool for improving team performance. The NTSB concludes that the captain’s indecision in responding to the anomaly and her failure to follow standard operating procedures was the result 65 NTSB Aircraft Accident Report 66 of a combination of poor CRM skills, limited experience as a pilot-in-command in the Learjet 60, and, during the accident sequence in particular, her less than confident and assertive leadership in the cockpit.

[ANALYSIS] Thus, one explanation for the captain’s incorrect briefing may be the use of automaticity in processing information, which is when a person uses low levels of attention to recite routine or habitual information. Such an error was consistent with a number of other information processing errors that the captain made during the taxi (including reading back wind 49 NTSB Aircraft Accident Report information incorrectly) that showed a lack of focus. Therefore, the NTSB concludes that there was no indication that the captain’s understanding of the rejected takeoff criteria was deficient; thus, the captain likely misspoke when she incorrectly stated the criteria in her pretakeoff briefing. During the takeoff, when the first tire failed and the rumbling noise began, the first officer stated, “go,” then “go, go, go.” The airplane’s ground speed at the time was about 137 kts, and, as shown by runway gouging and tire skid marks, the airplane veered to the right and across the runway centerline. Only debris from the right outboard tire was found at the runway location that coincided with the timing of this event; thus, the runway marks were likely created by the right outboard wheel rim contacting the runway surface and the skidding of the still-intact right inboard tire. (The airplane was initially left of the runway centerline before it veered.) Hydraulic fluid, consistent with that found in at least one severed brake hose, was present on some tire fragments. In the next second (2 seconds after the onset of the rumbling sound), the captain asked, “go?” At this point in the takeoff roll, the airplane neared its peak ground speed of about 144 kts (extrapolated data show that it may have reached about 150 kts within the next second) and began shedding fragments of a landing light and other pieces (which likely separated after having been impacted by fragments of the right outboard tire). The timing of the captain’s question to the first officer coincided with the captain reducing engine power for about 1 second, then increasing it for about 1 second before decreasing it again, about which time the first officer stated, “no? ar-right … what the [expletive] was that?” The entire RTO procedure, up to this point, spanned about 5 seconds since the onset of the rumbling noise from tire fragments, first from the right outboard tire and then from the right inboard tire. Although there is no indication that either the captain or the first officer knew what type of problem occurred, each reacted to it differently. The first officer’s statements to “go” suggest that, despite being unaware of the type of failure that occurred, he relied on his training and recognized that, once the airplane’s speed passes V1, the appropriate response is to continue the takeoff for nearly all anomalies except when airplane controllability is in serious doubt.

[ANALYSIS] Although there is no indication that either the captain or the first officer knew what type of problem occurred, each reacted to it differently. The first officer’s statements to “go” suggest that, despite being unaware of the type of failure that occurred, he relied on his training and recognized that, once the airplane’s speed passes V1, the appropriate response is to continue the takeoff for nearly all anomalies except when airplane controllability is in serious doubt. Both the captain and the first officer were trained that continuing the takeoff under such circumstances offers several safety advantages over an RTO, such as more time to analyze the situation, the ability to reduce the airplane’s gross weight and to use landing flaps, the ability to prepare for vibration and directional control problems on landing, and the availability of more runway on which to stop the airplane. The NTSB acknowledges that the first officer, as the pilot not flying, would not have received the same airplane controllability cues that the captain received, particularly when the airplane veered to the right after the first tire failure. Thus, the NTSB considered the possibility that the captain rejected the takeoff because of a perceived loss of airplane control. However, runway markings show that, after the airplane veered, it was realigned with the runway heading, indicating that the captain was able to regain and maintain directional control. Further, although the captain initially reduced engine power, she did not make an RTO callout, and her subsequent increase in engine power, concurrent with her question “go?,” indicates that she briefly considered continuing the takeoff before finally committing to the RTO. Therefore, the NTSB concludes that the captain’s uncertainty as to whether to continue the takeoff suggests that her 50 NTSB Aircraft Accident Report 51 initial action to reject it did not result from a perception that the airplane was uncontrollable and could not fly. During about 4 seconds after the captain made the second engine power reduction, the airplane’s engine N1 decreased to about 7,300 rpm, the captain made the comment “full out” (likely referring to full deployment of the thrust reversers), and wheel brakes were applied (as indicated by CVR sound evidence). Extrapolated ground speed information estimated that the airplane decelerated to about 128 kts. Debris evidence showed that, at this point, all of the MLG tires had failed (within about 9 seconds of the first tire’s failure). The captain’s action to reject the takeoff after the airplane had passed V1 placed the flight crew and passengers in a high-risk situation; accident and incident data show that high-speed RTOs can result in runway overruns. During pilot training, V1 is generally considered to be the speed at which the pilot must be committed to transition the airplane to flight; to reinforce this concept, pilot training programs and standard operating procedures (including those provided to the captain) emphasize that the pilot flying should remove his/her hand from the throttles at the time that the V1 callout is made.

[ANALYSIS] During pilot training, V1 is generally considered to be the speed at which the pilot must be committed to transition the airplane to flight; to reinforce this concept, pilot training programs and standard operating procedures (including those provided to the captain) emphasize that the pilot flying should remove his/her hand from the throttles at the time that the V1 callout is made. The purpose of such training and reinforcement is to ensure that the pilot responds immediately and correctly to any anomalies because the situation allows no time for assessment after the problem occurs, and any delays or mistakes increase the chance of an overrun (or the speed at which an overrun occurs). The NTSB concludes that, in the absence of evidence that the airplane was uncontrollable, the captain’s execution of a rejected takeoff for an unknown anomaly after the airplane’s speed had passed V1 was inconsistent with her training and standard operating procedures. 2.2.2 Uncommanded Forward Thrust Emergency As supported by the airplane performance study discussed in section 1.16.2, after the captain had committed to performing the high-speed RTO, the accident airplane’s thrust reverser system initially performed as commanded. On the basis of the captain’s comment “full out,” which coincided with a noticeable deceleration of the airplane, the study found that the thrust reversers had fully deployed, and the system provided reverse engine thrust. However, about 7 seconds after the captain committed (about 10 seconds after the rumbling noise began), the CAM captured the nosewheel steering disconnect warning tone. Because nosewheel steering is typically engaged while the airplane is on the ground, the timing of this tone provides an indication of when the system status changed to “air mode.” For this to occur, the circuit associated with the MLG electrical components and wiring, which include the wheel speed sensor and squat switch, must have sustained damage that affected the air-ground signal. Debris found on the runway and other physical evidence show that the MLG area where system components were mounted sustained damage from the shedding tire fragments.78 As a result, because the system logic requirements for maintaining thrust reverser deployment were no longer being met, the thrust reversers stowed. Meanwhile, as indicated by engineering and ground tests, the thrust reverser levers in the cockpit remained in the raised 78 The right landing light was mounted immediately above the squat switch, and the wiring was routed together. Glass from the light was found on the runway near the fragments from the right outboard and inboard tires. NTSB Aircraft Accident Report full-reverse-thrust position, the TR DEPLOY annunciators extinguished, and the TR UNLOCK annunciators momentarily illuminated (while the reverser doors were in transit toward stow); then, the TR ARM annunciators flashed briefly before all TR annunciators extinguished completely. During this sequence, the EECs shifted logic and signaled the FADEC components to change to the forward thrust power schedule.

[ANALYSIS] Therefore, the NTSB recommends that the FAA revise FAA Order 8110.48 to require that the most current airworthiness regulations related to equipment, systems, and installations (14 CFR 25.1309) are applied to all derivative design aircraft certificated as changed aeronautical products. The NTSB further recommends that the FAA review the designs of existing derivative design aircraft that were certificated as changed aeronautical products against the requirements of the current revision of 14 CFR 25.1309 and require modification of the equipment, systems, and installations to fully comply with this regulation. 2.4 Flight Crew Performance Section 2 of AC 120-62 acknowledges that tire failures may be difficult to identify from the flight deck and stresses that flight crews must be cautious not to inappropriately conclude that another problem exists. The accident airplane’s swerve, the onset of continuous noise from tire fragments striking the fuselage, and the related airframe vibration could have startled the captain. Further, the hydraulic fluid found on some tire fragments indicates that hydraulic integrity was compromised early in the sequence. As a result, the hydraulic pressure annunciators in the cockpit would have illuminated, providing the captain with additional cues about problems that she might not have fully comprehended. However, all of these cues occurred after the airplane had passed V1, and there was no strong evidence that the airplane was uncontrollable. The captain’s action to reject the takeoff and her lack of a callout, contrary to her training, may have been the result of the “startle factor,” which is often lacking in training scenarios. In most V1 training scenarios, pilots are in a simulator, are aware that they will be receiving an anomaly (usually an engine failure) on takeoff, and are prepared to respond. In the real world, the situation is more dynamic, the consequences are greater, and the pilot is not aware that a failure will occur or what type of failure it is. This “startle factor” can increase the stress level of the pilot, resulting in an incorrect decision being made. The following analysis examines factors that may have influenced the captain’s actions. 2.4.1 Lack of Training for Tire-Related Events As indicated in the NTSB’s 1990 SIR related to runway overruns associated with high-speed RTOs85 and an updated review by Boeing that includes data up to 1999, accidents and incidents related to RTOs initiated because of wheel or tire malfunctions are as common as those related to RTOs performed in response to engine failures. The accidents and incidents 85 NTSB SIR-90-02. NTSB Aircraft Accident Report show that, like the accident captain, many other pilots have misinterpreted tire anomalies and responded by initiating an unnecessary RTO after V1. According to an FSI instructor, the training curriculum provided to the captain and the first officer did not include any scenarios in which a tire failure occurred.

[ANALYSIS] Less than 2 seconds later, however, when the airplane was more than 2,500 feet down the runway (with about 6,100 feet remaining), the CVR captured the beginning of a loud rumbling noise. The airplane’s location on the runway at the onset of the noise correlated with the location where the first pieces of right outboard MLG tire were found. Thus, the onset of the loud rumbling noise likely resulted from pieces of the right outboard tire separating from the wheel and striking the underside of the airplane and was likely accompanied by shaking and vibration of the airframe. From this point forward, the accident sequence can be divided into two distinct segments. The first segment involves the captain’s initiation of the high-speed RTO, which was a high-risk event. The second segment of the accident sequence involves the uncommanded forward thrust emergency related to the uncommanded stowage of the airplane’s thrust reversers. 2.2.1 Captain’s Initiation of Rejected Takeoff After V1 The captain and the first officer were trained that rejecting a takeoff is acceptable for any anomaly occurring before the airplane reaches 80 kts and that, for speeds between 80 kts and V1, the takeoff could be rejected for major anomalies, such as catastrophic failure, engine fire, engine failure, thrust reverser deployment, or loss of directional control. Their training and standard operating procedures indicated that, because of the high risk of runway overrun and other dangers, rejecting a takeoff at speeds greater than V1 should be performed only when airplane control is seriously in doubt. During the captain’s pretakeoff briefing, she incorrectly stated that the takeoff could be rejected for major anomalies occurring between V1 and V2. None of the captain’s training provided by FSI and Global Exec Aviation referenced any RTO criteria beyond V1, and no evidence indicated that the captain was unfamiliar with the concept of V1 as it related to go/no-go decision-making. An instructor who provided the captain with recurrent ground and simulator training, which included V1 cuts and RTOs, described her as meticulous with good organizational skills. As noted previously, V1 for the accident flight conditions was about 136 KIAS. Given that V2 (which is a takeoff safety speed that would provide a minimum climb gradient after a loss of engine power [about 153 KIAS]) occurs after Vr (which is the airplane’s rotation speed [about 145 KIAS]), it is unlikely that the captain would have considered V2 to be part of the RTO criteria. A pretakeoff briefing is fairly standard in that the RTO criteria rarely change from one takeoff to the next. Thus, one explanation for the captain’s incorrect briefing may be the use of automaticity in processing information, which is when a person uses low levels of attention to recite routine or habitual information. Such an error was consistent with a number of other information processing errors that the captain made during the taxi (including reading back wind 49 NTSB Aircraft Accident Report information incorrectly) that showed a lack of focus.

[ANALYSIS] Also, her attention was divided between crosschecking the takeoff performance numbers and visually clearing the runway, reducing the amount of time she devoted to inspecting the external environment. It is likely that, due to her divided attention, she performed a cursory visual scan of the runway environment and made the associated routine callout. Thus, workload, multitasking, time pressure, and expectancy likely played a role in the FO not detecting the captain’s surface navigation errors between the time the airplane passed taxiway K and the time the airplane taxied onto runway 4L. The FB recalled that his visual attention was focused inside the flight deck from the time the crew received the load closeout until the airplane crossed the holding position markings and entered runway 4L. During that period, he recalled handing the printed load closeout to the other crewmembers, cleaning up paperwork on the center pedestal, and adjusting his radios. He recalled hearing some confusion about crossing a runway and looking up, at which time the airplane was across the runway 4L hold short markings. As a result, he would no longer have been able to see the signs for runway 4L. Thus, the FB did not detect the captain’s deviation from AAL106’s taxi clearance because he was performing operational tasks other than looking outside and monitoring the airplane’s position. Aviation Investigation Report AIR-24-01 39 The NTSB concludes that the FO and FB were likely distracted from their primary duty of assisting the captain in safely taxiing the airplane by other operational activities, which resulted in the crew’s loss of situational awareness during a critical phase of flight. 2.2.1.3 American Airlines Crew’s Nondetection of DAL1943 on Runway 4L DAL1943 was positioned for takeoff on runway 4L for about 2 minutes before beginning its takeoff roll, and began its takeoff roll about the time AAL106 started turning onto taxiway J. As discussed in the previous section, the AAL106 captain and FO recalled visually scanning the runway, but neither recalled seeing DAL1943 on the runway. The captain and FO recalled scanning the runway at some point during the 11 seconds between the time the captain began turning the airplane onto taxiway J and the time the airplane crossed the hold short markings and the RELs activated. The crew recalled that the captain also switched on some additional exterior lights during this time, which briefly required some of his visual attention inside the cockpit. As the airplane began to turn onto and align with taxiway J, the approach end of runway 4L was to the captain’s right and may have been partially obstructed by the FO’s body and aircraft structure. Thus, it is possible the captain would not have had an opportunity to see DAL1943 due to his obstructed view. Examination of satellite images, illumination diagrams, and on-scene nighttime photographs of the airport suggest that a significant amount of airport lighting was present between the two airplanes in various colors and intensities.

[ANALYSIS] Believing that AAL106 was conducting the taxi as instructed, he did not ensure that the flight crew completed the right turn onto taxiway K before he began attending to other duties. FAA Order 7110.65Z, Air Traffic Control, paragraph 2−1−2, Duty Priority, stated that controllers were to give first priority to separating aircraft and issuing safety alerts as required. Therefore, the NTSB concludes that the ground controller expected the AAL106 flight crew to adhere to the assigned taxi instructions and did not detect the flight crew’s surface navigation error and subsequent turn onto taxiway J because he was performing a lesser priority task that entailed looking down. 2.5.2 Local Controller The local controller stated that he performed a visual scan before issuing the takeoff clearance to DAL1943, then he started talking to other airplanes that were taxiing to line up and wait, one on runway 4L and another on runway 31L. Aviation Investigation Report AIR-24-01 50 Shortly after the controller provided these instructions, the ASDE-X alerted. He looked at the ASDE-X display and saw the aircraft involved were DAL1943 and AAL106. The local controller cancelled DAL1943’s takeoff clearance 5 seconds after the ASDE-X alerted. He repeated the instruction a second time, and the DAL1943 crew responded that they were rejecting the takeoff. The NTSB concludes that the local controller acted in a timely and appropriate manner following the ASDE-X alert by cancelling DAL1943’s takeoff clearance. 2.5.3 Air Traffic Control Tower Team At the time of the incident, the tower was changing the runway configuration, which required some of the controllers to physically move into different positions in the tower, as well as change all their equipment to the new configuration. Airport operations was on the airfield conducting checks and switching equipment for the runway change as well. On the night of the incident, the runway configuration change was requested to be completed earlier than the time it typically took place between 2130 and 2230. Although a runway configuration change is an important priority, and, on the night of the incident, occurred at an atypical time, it is a routine event that does not always occur at prescribed times. The whole tower team has the primary responsibility of ensuring safe and effective operations in the tower and to perform continual visual scans of the runways. In this case, the cab coordinator, flight data/clearance delivery controller, and operations supervisor were busy making preparations for the runway change and were not giving their full attention to the air traffic operations at the time. The tower team, collectively, did not effectively prioritize their duties by ensuring that the team maintained a continuous scan of the airport environment as preparations for the runway change were performed.

[ANALYSIS] In postincident interviews, all three crewmembers described an increasing workload as the taxi progressed. The captain attributed his error, in part, to this factor. The instruction provided to the crew to cross runway 31L on taxiway K could have served as a salient cue reminding the captain to turn right at taxiway K if it had been issued closer to that intersection; however, because this clearance was issued 4 minutes before, visual indications were the primary cues of the airplane’s position as the airplane approached taxiway K. The night conditions at the time of the incident would have reduced positional cues. The airplane had passed through 17 taxiway intersections before reaching taxiway K, so the mere presence of an intersection was not sufficient to indicate arrival at a decision point. There was a surface painted taxiway K sign and an arrow pointing toward taxiway K located just before the intersection. This sign would have been visible for at least a few seconds when illuminated by the airplane’s taxi lights. There was also a lighted yellow sign on the right side of taxiway K labeled “4L,” which pointed the way to that runway. This sign would have been viewable as the airplane approached the intersection with taxiway K; however, it did not explicitly label taxiway K. Thus, two cues were available, but one was only temporarily visible, and the other required inference to determine that it denoted the intersection of taxiways B and K. ACARS records indicated that the crew accepted the airplane’s zero fuel weight and center of gravity in the FMC about 30 seconds before the airplane crossed taxiway K. The receipt of this information led to a cascade of other crew activities as the captain and FO reviewed and crosschecked takeoff performance calculations. These tasks, triggered by the unpredictable timing of the load closeout receipt, required at least some of the captain’s visual attention and likely reduced his attention to and visual monitoring of the external environment. Thus, although salient cues to the airplane’s position were present and could have reminded the captain of the need to turn right at taxiway K, they were not sufficient to break through the distraction caused by the crew’s load closeout-related tasks. 27 The expected tasks included obtaining the load closeout, performing follow-on tasks related to the load closeout, communicating with ATC, addressing the passengers about the departure, communicating with the cabin crewmembers, and performing the before takeoff checklist. The unexpected tasks included dealing with multiple weather advisories from dispatch via ACARS and informing crewmembers accordingly. The tasks with timing determined by external events included those tasks triggered by the receipt of weather advisories and tasks triggered by receipt of the load closeout, such as crosschecking final weights and V-speeds.

[ANALYSIS] The unexpected tasks included dealing with multiple weather advisories from dispatch via ACARS and informing crewmembers accordingly. The tasks with timing determined by external events included those tasks triggered by the receipt of weather advisories and tasks triggered by receipt of the load closeout, such as crosschecking final weights and V-speeds. Aviation Investigation Report AIR-24-01 37 An alternative framework from the scientific literature on human error involves branching error, which can occur when a task involves an action schema that is associated with more than one possible end goal. In such situations, periodic attentional checks are needed to detect one’s arrival at a choice point. This is consistent with the previous discussion about the captain’s prospective memory error. In a branching error, this check does not occur, behavior becomes disconnected from intention, and actions beyond the choice point become driven by the schemas that are commonly performed in the new situational context. This latter form of a skill-based action slip is known as environmental capture (Reason 1990). According to this framework, lower-level schemas caused the captain to continue the taxi along taxiway B after he missed the turn at taxiway K. Shortly thereafter, the airplane was directly facing taxiway J and the highly salient flashing yellow guard lights for runway 4L. A typical behavior pattern performed in that location involved taxiing onto taxiway J and crossing runway 4L (the standard route to runway 31L). In the absence of focused attention, schema theory suggests that the captain’s actions may have become disconnected from his intentions and situational cues could have prompted him to continue along a common, but incorrect, route. This would explain the captain’s subsequent surprise and confusion when the ground controller alerted him to the discrepancy between AAL106’s clearance and the airplane’s position. Thus, a branching error appears to be a good fit for the captain’s continuation of the taxi onto taxiway J and across runway 4L. The NTSB concludes that the captain’s deviation from AAL106’s taxi clearance likely resulted from several factors, including an early clearance to cross runway 31L, interruptions and multitasking related to the crew’s delayed receipt of the load closeout, the captain’s prospective memory error in forgetting to turn right at taxiway K, and environmental capture, which prompted the captain to proceed along a familiar, but incorrect, route. 2.2.1.2 American Airlines FO and FB’s Nondetection of the Captain’s Error The NTSB sought to understand why neither the FO nor the FB detected the captain’s error between the time the airplane passed taxiway K until the RELs activated. The FO told investigators she was head-down dealing with tasks related to the load closeout for the approximate 1.5 minutes from the time the load closeout was received until the captain began turning onto taxiway J and toward runway 4L. The missed turn at taxiway K occurred during this time that the FO’s visual attention was likely focused inside the flight deck.

[CONCLUSIONS > FINDINGS] Kennedy International Airport was operational at the time of the incident and functioned as designed, generating both aural and visual alerts when American Airlines flight 106 crossed runway 4L while Delta Air Lines flight 1943 was departing, and likely reduced the severity of the incident by preventing a runway collision. 10. The ground controller expected the American Airlines flight 106 crew to adhere to the assigned taxi instructions and did not detect the flight crew’s surface navigation error and subsequent turn onto taxiway J because he was performing a lesser priority task that entailed looking down. 11. The local controller acted in a timely and appropriate manner following the airport surface detection equipment model-X alert by cancelling Delta Air Lines flight 1943’s takeoff clearance. 12. The John F. Kennedy International Airport air traffic control tower team had the responsibility of scanning the runways and airport environment but did not effectively prioritize their duties to ensure a continuous scan, which resulted in their nondetection of the American Airlines flight 106 crew’s deviation from taxi instructions. 13. Cockpit voice recorders (CVRs) with a 25-hour recording capability are necessary because valuable information continues to be overwritten on CVRs that are designed to record only 2 hours of audio data. 3.2 Probable Cause The NTSB determines that the probable cause of this incident was the American Airlines flight 106 (AAL106) crew’s surface navigation error due to distractions caused by their performance of concurrent operational tasks during taxi, which resulted in a loss of situational awareness. Contributing to the incident was the air traffic control tower team’s nondetection of the AAL106 crew’s deviation from taxi instructions while performing concurrent operational tasks; the timing of the runway status light system, which activated too late to prevent the AAL106 crew from crossing the runway hold short line; and American Airlines’ lack of adequate risk controls to prevent concurrent flight crew tasks from leading to distraction, loss of situational awareness, and deviation from an authorized taxi clearance. Reducing the severity of the incident, and likely preventing an accident, was the activation of the ASDE-X warning in the air traffic control tower and the local controller’s prompt cancellation of DAL1943’s takeoff clearance. Aviation Investigation Report AIR-24-01 54

[ANALYSIS] Each crewmember reported being rested and was observed to be alert on the evening of the incident. Aviation Investigation Report AIR-24-01 34 • Air traffic control tower staffing: JFK tower was staffed with seven controllers staffing the eight core control positions. Some positions were combined due to the lower traffic level at the time of the incident. This staffing was normal and adequate for the time of night, complexity, and volume of traffic. Thus, the NTSB concludes that none of the following were factors in this incident: (1) pilot and controller qualifications, (2) flight crew fatigue, and (3) ATCT staffing. 2.2 Incident Sequence On the evening of the incident, the flight crew of AAL106 was preparing to taxi from gate 18 at JFK to runway 31L. Most of the flight’s paperwork, as well as the initial taxi briefing conducted by the flight crew, anticipated a departure from runway 31L; however, upon contacting the ground controller following pushback, the crew received instructions to taxi to runway 4L. The captain and FO then conducted a revised briefing for a taxi to runway 4L. 2.2.1 American Airlines Flight Crew’s Deviation from Taxi Clearance 2.2.1.1 Prospective Memory Errors In the postincident interview, the FO and captain stated that the initial taxi route briefing for the anticipated departure from runway 31L included taxiway TA, left on B, and hold short at J, indicating that the crew would receive a subsequent clearance to proceed onto taxiway J and cross runway 4L on the way to runway 31L. The captain stated that this was the standard route from the American Airlines gates to runway 31L. The FO recalled that, after learning the flight would depart from runway 4L, the captain’s revised taxi briefing stated, “Same taxi route up until Bravo short of Kilo.” The captain recalled briefing the FO and FB that the controller would “have you come up Kilo, and then as you get closer, they’ll tell you when to cross,” implying two segments to the taxi route, with two associated clearances from the ground controller. The captain stated that this was the typical route from the American Airlines gates to runway 4L. Human factors research indicates that, through extensive practice, experts learn to perform routine tasks, like taxiing an airplane, automatically (Wickens & McCarley 2008). This reduces attentional requirements and frees cognitive resources for other tasks. Psychological theory suggests that automatic performance is governed by sensory-motor knowledge structures known as schemas (Cooper 2016). When a goal is identified, a parent schema is established (Norman 1981). When Aviation Investigation Report AIR-24-01 35 taxiing an airplane, a pilot’s intended destination can be considered a parent schema. Subordinate schemas would address segments of the route, and lower-level schemas would govern psychomotor tasks, like controlling speed and steering the airplane along the taxiway centerline.

[ANALYSIS] The FO told investigators she was head-down dealing with tasks related to the load closeout for the approximate 1.5 minutes from the time the load closeout was received until the captain began turning onto taxiway J and toward runway 4L. The missed turn at taxiway K occurred during this time that the FO’s visual attention was likely focused inside the flight deck. The FO recalled hearing the captain call out, “cleared to cross” as the captain was turning the airplane toward a runway (onto taxiway J). Based on ADS-B data, this turn occurred about 2044:29. Upon hearing the captain’s callout, she looked up and visually inspected the runway as the captain did the same, in accordance with the AAL Aviation Investigation Report AIR-24-01 38 FOM. The captain and FO then made the callouts “clear left” and “clear right,” respectively. Both pilots’ interviews indicated that this visual inspection was completed before the RELs illuminated (at 2044:40). Thus, the FO’s visual inspection occurred between 2044:29 to 2044:40, a period of about 11 seconds, while the airplane was traveling at a speed of 11 kts toward the runway. Two types of runway signs were available in the FO’s field of view when she scanned the runway: a pair of white-on-red signs painted on the surface of taxiway J before the holding position markings, and a pair of vertical black-on-yellow lighted signs erected on either side of the runway hold short position markings. The painted signs were likely viewable for about 4 seconds and the lighted signs were likely viewable for an overlapping period of 8 seconds. One or the other type of sign was likely viewable for a period of about 10 seconds. Ten seconds is sufficient for locating and reading approaching runway signs if a viewer is specifically trying to do so; however, the FO either did not read the signs or did not notice that the numbers on the runway signs conflicted with the AAL106’s taxi clearance and that the airplane was about to cross the wrong runway. Psychological research indicates that people only perceive and remember fine details of a visual scene that receive focused attention. It is likely that the FO did not notice the lettering on the signs because she did not focus visual attention on the signs. Visual attention is highly influenced by expectancy (Downing 1988). The FO knew the captain was very familiar with JFK; therefore, she likely expected that the odds of his making a surface navigation error were quite low. Also, her attention was divided between crosschecking the takeoff performance numbers and visually clearing the runway, reducing the amount of time she devoted to inspecting the external environment. It is likely that, due to her divided attention, she performed a cursory visual scan of the runway environment and made the associated routine callout.

[ANALYSIS] Therefore, the NTSB reiterates Safety Recommendation A-18-30. Because the FAA did not complete responsive action to Safety Recommendation A-18-31 by January 1, 2024, this recommendation is classified Closed—Unacceptable Action/Superseded. The NTSB recommends that the FAA require retrofit of all CVRs on all airplanes required to carry both a CVR and a flight data recorder with a CVR capable of recording the last 25 hours of audio. Safety Recommendation A-24-9 is classified Open—Unacceptable Response. Aviation Investigation Report AIR-24-01 52 3. Conclusions 3.1 Findings 1. None of the following were factors in this incident: (1) pilot and controller qualifications, (2) flight crew fatigue, and (3) airport traffic control tower staffing. 2. The captain’s deviation from American Airlines flight 106’s taxi clearance likely resulted from several factors, including an early clearance to cross runway 31L, interruptions and multitasking related to the crew’s delayed receipt of the load closeout, the captain’s prospective memory error in forgetting to turn right at taxiway K, and environmental capture, which prompted the captain to proceed along a familiar, but incorrect, route. 3. The first officer and relief first officer were likely distracted from their primary duty of assisting the captain in safely taxiing the airplane by other operational activities, which resulted in the crew’s loss of situational awareness during a critical phase of flight. 4. The American Airlines flight crew’s nondetection of Delta Air Lines flight 1943 (DAL1943) on runway 4L likely resulted from night conditions, the location of DAL1943 within a complex array of airport lights, the distance between the two airplanes, the lack of relative motion of DAL1943 in the visual field of the crew of American Airlines flight 106, and expectation bias. 5. A procedural crosscheck that requires a flight crew to verbalize the number of a runway they are about to cross, as indicated by runway signs, would reduce the likelihood of future runway incursions resulting from flight crew surface navigation errors. 6. Additional risk mitigation strategies as part of an operator’s safety management system would reduce the likelihood that flight crew performance of concurrent tasks during taxi will lead to inaccurate navigation on the airport by reducing distractions associated with multitasking. 7. The implementation of a flight deck alerting system on transport-category aircraft that provides alerts of traffic on a runway or taxiway and traffic on approach to land would enhance safety by providing pilots with improved situational awareness and would reduce the risk of future runway-related incidents and accidents. 8. The runway entrance lights operated as designed; however, they were ineffective in preventing the crew of American Airlines flight 106 from crossing Aviation Investigation Report AIR-24-01 53 the runway 4L hold short markings because they activated too late for the crew to perceive them and stop the airplane in a safe area. 9. The airport surface detection equipment model-X system at John F.

[ANALYSIS] As a result, an aural “runway occupied” alert was voiced, and the corresponding visual text alert “RWY 4LIDAL1943, AAL106IRWY OCCUPIED” was displayed on the air traffic displays; the ASDE-X system functioned as designed. The local controller cancelled DAL1943’s takeoff clearance 5 seconds after the ASDE-X aural alert was annunciated. The NTSB concludes that the ASDE-X system at JFK was operational at the time of the incident and functioned as designed, generating both aural and visual alerts when AAL106 crossed runway 4L while DAL1943 was departing, and likely reduced the severity of the incident by preventing a runway collision. 29 While paperwork obtained from the Port Authority of NY and NJ dated January 15, 2023, identified RELs 3 and 4 as being inoperative, ASDE-X playback showed the inoperative fixtures were RELs 2 and 3. Aviation Investigation Report AIR-24-01 49 2.5 Air Traffic Control 2.5.1 Ground Controller The ground controller instructed the crew of AAL106, from their position at taxiway TA after pushing back from the gate, to taxi to runway 4L, via a left turn onto taxiway B, and to hold short of taxiway K. As AAL106 continued to taxi on taxiway B, the ground controller instructed the crew to cross runway 31L at taxiway K. The crew read back the instruction correctly, which likely reinforced the ground controller’s belief that the crew understood the taxi instructions. The ground controller later reported in a postincident interview that he observed AAL106 swing out to the left on taxiway B to begin what he believed to be a typical wide right turn onto taxiway K. Radio communications showed that, as AAL106 passed taxiway K, the ground controller received a request from, and subsequently provided taxi instructions to, another airplane. He also established communications with an airport operations vehicle and requested that they sweep runway 31R in preparation for the runway configuration change. The ground controller recalled that, after he finished his communication with the airport operations vehicle, he looked down to attend to another task. While the ground controller was looking down, AAL106 continued the slight left turn on taxiway B, then turned right onto taxiway J, where its nose would eventually cross active runway 4L, triggering an ASDE-X alert. The ground controller stated that he expected the flight crew of AAL106 to adhere to their taxi instructions given that they correctly read back the instructions as issued. Believing that AAL106 was conducting the taxi as instructed, he did not ensure that the flight crew completed the right turn onto taxiway K before he began attending to other duties. FAA Order 7110.65Z, Air Traffic Control, paragraph 2−1−2, Duty Priority, stated that controllers were to give first priority to separating aircraft and issuing safety alerts as required.

[ANALYSIS] In the postincident interview, the captain said that he did not normally need to perform these tasks during taxi because load closeouts typically arrive soon after the airplane pushes back from the gate. He said that he normally made a point of crosschecking the load closeout with the FO and completing related tasks before departing the ramp area. In this case, however, he said that he decided to begin the taxi before receiving the load closeout because AAL106 had a third flight crewmember (the FB). He explained that many of the flights he operated from JFK to London had only two flight crewmembers, and he believed the presence of the FB would allow the crew to handle more tasks during the taxi while maintaining a manageable workload. The captain delegated to the FB the duty of following up with dispatch to ensure that the load closeout was received; however, the captain could not delegate crosschecking the takeoff performance numbers with the FO. As previously discussed, load closeout-related tasks subsequently played a role in the captain’s distraction and surface navigation error. The American Airlines FOM stated that flight crewmembers should use a continuous loop process for actively monitoring and updating their progress and location during taxi. This includes knowing the aircraft’s present location and mentally calculating the next location on the taxi route that will require increased attention. Although monitoring is important, a pilot cannot continuously focus on monitoring one task while simultaneously engaging in another concurrent task, particularly one that shares the same sensory channel (that is, vision). As previously discussed, this requires multitasking, which can lead to errors. The FOM also stated that other flight deck duties and non-ATC communications should not divert attention from the safe movement of the aircraft, especially at critical times, such as runway crossings and transitioning through complex taxiway intersections. However, any time concurrent tasks are performed while taxiing, some conscious attention is diverted away from the surface navigation task. Aviation Investigation Report AIR-24-01 44 Airline crews strive for efficiency and maintaining on-time performance; therefore, it is understandable that the crew of AAL106 opted to perform load closeout tasks during taxi. Doing so was not prohibited because these tasks are essential to the safety of the flight, and the crew likely believed it would help them avoid a delay. If American Airlines guidance had specified that the captain and FO had to complete load closeout tasks before beginning the taxi, the crew likely would have completed these tasks in the ramp area. Alternatively, if such guidance stated that load closeout-related tasks could only be performed when the airplane was parked, the crew might have delayed action on those tasks until they reached runway 4L for takeoff and parked the airplane, thereby increasing their external monitoring while the airplane was moving. Thus, a lack of guidance prohibiting the performance of load closeout related tasks during taxi increases the risk that a similar surface navigation error may occur in the future. This incident demonstrates how interruptions and multitasking can reduce crew awareness and increase the risk of surface navigation errors, runway incursions, and accidents.

[ANALYSIS] Subordinate schemas would address segments of the route, and lower-level schemas would govern psychomotor tasks, like controlling speed and steering the airplane along the taxiway centerline. Although schema-driven action requires little conscious attention, attention is required at key points to ensure that appropriate subordinate or lower-level schemas are activated at appropriate times. In this case, the captain’s goal was to taxi to runway 4L along the typical route, which was highly familiar to him. After AAL106 departed the gate, the taxi task required little of the captain’s conscious attention for several minutes; the captain merely had to steer the airplane along the lighted centerline of taxiway B and maintain an appropriate speed. When the airplane reached taxiway K, however, the captain needed to bring his attention back to the taxi task, verify that AAL106 had received a clearance to cross runway 31L, turn right on taxiway K, and cross runway 31L. The crew received a clearance to cross runway 31L when the airplane was about 4 minutes from arriving at taxiway K. The FO acknowledged the clearance on the radio. The captain told investigators he believed he heard and understood this clearance. Thus, he knew that he no longer needed to hold short of taxiway K and await clearance to turn right on taxiway K; however, he still needed to remember to turn right at that location where the standard routes to runways 4L and 31L diverged. Psychologists use the term prospective memory to refer to the set of cognitive processes that govern the maintenance, retrieval, and execution of deferred tasks (Dismukes 2012). The captain’s omission of the right turn at taxiway K, despite his intention to do so, was a prospective memory error. A delay between the formation of an intention to do something and the execution of that task, which occurred in this case, increases the likelihood of a prospective memory error. This occurs because intentions may be overwritten by the cognitive demands of intervening tasks (Wilson et al. 2020). Factors associated with prospective memory errors include interruptions, multitasking, and the absence of salient cues that would normally prompt the performance of a habitual task. (Loukopoulos, Dismukes, and Barshi 2009). The NTSB explored the possible role of each in the incident. The AAL106 crew was performing multiple tasks during the taxi. The timing of some of these tasks was determined by the crew, but the timing of others was Aviation Investigation Report AIR-24-01 36 determined by external events.27 This mix of tasks with expected and unexpected timing led to interruptions and reprioritization and interleaving of tasks, otherwise known as multitasking. In postincident interviews, all three crewmembers described an increasing workload as the taxi progressed. The captain attributed his error, in part, to this factor.

[ANALYSIS] As a result, the captain’s troubleshooting during the accident sequence was confused. According to the NTSB’s performance study, the lateral and directional handling problems that would result from inadvertent autopilot activation and mistrim under these circumstances would result in control forces that were light at first and larger as the airplane accelerated and turned further from the runway extended centerline heading. The effects 75 The Citation 550 pilots involved with the simulator testing during the investigation stated that each, at one time in their careers, had inadvertently pushed the autopilot instead of the yaw damper buttons. 76 About 35 seconds after takeoff, after the captain made his third mention of an unspecified control problem, the first officer stated, “how’s your trim set? Is that the way you want it?” About 12 seconds later, after the captain stated that the airplane wanted to “turn hard left,” the first officer asked, “how’s your trim down here?” Finally, about 1 ½ minutes after takeoff, the first officer asked, “how’s that, any better?” NTSB Aircraft Accident Report 45 of the autopilot turning the airplane toward the runway’s magnetic heading and the forces resulting from the mistrimmed condition ultimately would have demanded strenuous inputs from both pilots to maintain control. The performance study results were consistent with many aspects of this interpretation of the accident sequence. For example, the captain first complained of a control problem just moments after the airplane would have begun to turn from the initial autopilot engagement heading in accordance with the departure clearance, when an inadvertently activated autopilot would have been trying to return the airplane to the initial autopilot engagement heading. About 1 1/2 minutes later, after the control problem had escalated (likely as a result of the first officer’s trim inputs and the captain allowing the airplane to accelerate while he dealt with the unidentified control problem), the captain decided to declare an emergency and return to MKE. The final loss of control began about 1 minute later, just after the captain transferred the controls to the first officer to look for the autopilot circuit breaker. The performance study showed that the transfer of control occurred when the airplane was heading back towards MKE and passed through and moved away from the autopilot engagement heading, which, according to this scenario, would have resulted in dynamically changing forces on the control wheel as the autopilot reversed the direction of its input in an attempt to return to the runway heading. Other evidence was inconsistent with this scenario, however. For example, examination of the recovered autopilot servomotor clutches and the unstretched filaments in the cockpit autopilot-engaged light bulb (which was not illuminated at impact) suggested that, although the autopilot did have power at impact, it was likely not engaged at impact.77 Some stretching was observed in the yaw-damper-engaged bulb filaments.

[ANALYSIS] Because the CVR recorded no pitch-trim-related complaints by the pilots during their approach to MKE, it is likely that this short circuit occurred after the pilots’ last use of the electric pitch trim during the airplane’s landing at MKE. Examination of the wreckage also revealed that a contact in one of the two switches in the first officer’s pitch-trim control was bent inward, eliminating the space between it and the center reed. In normal pitch trim operations, the center reed moves outward and contacts the fixed contact to complete the ground path circuit; this condition might have provided the pitch trim motor with the nose-down latent ground path. (Impact damage precluded positive identification of the ground path.) Other evidence is inconsistent with this scenario. For example, previous NTSB investigations involving pitch control problems81 indicate that pilots are typically immediately aware of the nature of pitch-related problems and struggle to keep the airplane upright.82 Because the captain would have routinely adjusted the pitch trim with each configuration change and in response to any unwanted trim forces experienced during departure, it is likely that he would have recognized a runaway pitch trim situation shortly after it began. Further, in this scenario, the pilots would have been struggling to keep the airplane from nosing over during the accident flight, yet there was no specific reference to “pitch” or “pull” by either pilot until about 5 seconds before impact. 79 Evidence and maintenance records showed at least two instances in which the wires in the first officer’s control yoke/column had been previously repaired. 80 There was also evidence of a short circuit on the power wire for the autopilot clutches. 81 For example, see (a) Loss of Control and Impact with Pacific Ocean, Alaska Airlines Flight 261, McDonnell Douglas MD-83, N963AS, About 2.7 miles north of Anacapa Island, California, January 31, 2000, Aircraft Accident Report NTSB/AAR-02/01, (Washington, DC: National Transportation Safety Board, 2002); (b) Loss of Control on Takeoff, Emery Worldwide Airlines, Flight 17, McDonnell Douglas DC-8-71F, N8079U, Rancho Cordova, California, February 16, 2000, NTSB/AAR-03/02 (National Transportation Safety Board: Washington, DC, 2003); (c) Colgan Air (dba USAirways), Beechcraft 1900D, N240CJ, Yarmouth Massachusetts, August 26, 2003, NTSB/AAR-04/01 (National Transportation Safety Board: DC, 2001); and (d) Loss of Pitch Control During Takeoff, Air Midwest Flight 5481, Raytheon (Beechcraft) 1900D, N233YV, Charlotte, North Carolina, January 8, 2003, Aircraft Accident Report NTSB/AAR-04-01 (National Transportation Safety Board: DC, 2004); (e) The factual report and probable cause statement for a September 1, 2008, accident (NTSB identification number CHI08MA270), available on the NTSB website at <http://www.ntsb.gov/ntsb/query.asp>. 82 In addition, during investigation-related simulation sessions, pilots indicated that a natural first response to a pitch-related problem would be to press the autopilot/trim disconnect button.

[ANALYSIS] The captain repeatedly stated that he was “fighting the controls” (about 1557:51; 1557:57; 1558:07) and confirmed with the first officer that the landing gear had been retracted (about 1558:03). About 1558:21, the captain again stated that he was “fighting the controls,” this time adding, “It wants to turn hard left.” The captain allowed the airplane to accelerate and climb after takeoff consistent with the airplane’s assigned departure heading and altitude while he and the first officer tried to troubleshoot the control anomaly. Comments recorded by the CVR indicated that the pilots did 42 NTSB Aircraft Accident Report 43 not consult an emergency or abnormal procedure checklist.72 Further, CVR evidence indicated that, on at least one occasion, it is likely that the first officer made an unrequested trim input. About 1559, at the captain’s request, the first officer advised MKE air traffic controllers that they had a control problem and would be returning to the airport. For the remainder of the flight, the CVR recorded both pilots communicating on the radio. The CVR also recorded several instances in which the captain asked the first officer to help him maintain control of the airplane; the last such instance occurred about 1600:36, when the captain asked the first officer to hold the airplane’s flight controls so that he could try to pull circuit breakers. The last radar return was recorded moments later (just before 1601), about 8 to 9 seconds before the airplane hit the water in a steep nose-down, left-wing-low attitude. 2.2.1 Possible Initiating Event Accident Scenarios The NTSB evaluated several possible explanations for the initiating event experienced by the pilots and found that two of those possible explanations were most consistent with the abnormal control situation: 1) an inadvertent autopilot engagement, or 2) a runaway electric pitch trim. Examination of the available evidence, including the items described below, both supports and contradicts these scenarios. Without an FDR or image recorder on board, it was not possible to determine the exact cause of the initiating event or the pilots’ actions during the accident sequence. If the accident airplane had been equipped with crash-protected data- or image-recording equipment,73 the NTSB would also have had useful information regarding many other unresolved issues in this accident, including whether checklists were silently followed (partially or completely), whether and when the trim controls were used, and any other actions and flight control movements made by the pilots during the event.

[CONCLUSIONS > FINDINGS] Upon receipt of such information, inspectors should increase their oversight of operators who appear to be in financial distress. 60 NTSB Aircraft Accident Report 3. Conclusions 3.1 Findings 1. The captain and first officer were properly certificated and qualified under federal regulations to act in their respective roles during the accident flight. There was no evidence of any medical conditions that might have adversely affected the pilots’ performance during the accident flight. 2. Although the captain’s pilot certificates had previously been revoked because of a felony conviction involving the illegal transport of drugs into the United States, the Federal Aviation Administration had reissued his pilot certificates, and they were valid at the time of the accident. 3. The accident airplane was properly certificated, was equipped and maintained in accordance with industry practices (except for the wiring installed in the pilots’ control yokes), and was within weight and center of gravity limits. 4. If the accident airplane had been equipped with a recorder system that captured cockpit images and parametric data, investigators would have been better able to determine the circumstances that led to this accident. 5. The accident sequence initiated as a result of a control problem that was related to either an inadvertent autopilot activation or a pitch trim anomaly, the effects of which were compounded by aileron and/or rudder trim inputs; however, it was not possible to determine the exact nature of the initiating event. 6. Regardless of the initiating event, if the pilots had simply maintained a reduced airspeed while they responded to the situation, the aerodynamic forces on the airplane would not have increased significantly; at reduced airspeeds, the pilots should have been able to maintain control of the airplane long enough to either successfully troubleshoot and resolve the problem or return safely to the airport. 7. Pilots would benefit from training and readily accessible guidance indicating that, when confronted with abnormal flight control forces, they should prioritize airplane control (airspeed, attitude, and configuration) before attempting to identify and eliminate the cause of the flight control problem. 8. The design and location of the yaw damper and autopilot switches on Cessna Citation series airplanes do not adequately protect against inadvertent activation of a system, which could have disastrous consequences. 61 NTSB Aircraft Accident Report 9. A rounded type of sheathed wire bundle would fit better and be better protected within the control column shaft than the currently installed flat ribbon cable; replacement of the flat ribbon cable with a rounded type of sheathed wire bundle could result in fewer short circuits and other electrical events. 10. The incorporation of an aural pitch trim-in-motion warning and contrasting color bands on the pitch trim wheel in all Cessna Citation series airplanes would help pilots of those airplanes to more promptly recognize and correct runaway pitch trim situations before control forces become unmanageable. 11.

[ANALYSIS] No autopilot disconnect chime was recorded by the CVR. NTSB Aircraft Accident Report 47 2.2.1.3 Accident Scenario Analysis and Discussion Although these two scenarios were the most likely explanations for the initiating flight control event, neither scenario was completely consistent with the investigative evidence. Therefore, the NTSB concludes that the accident sequence initiated as a result of a control problem that was related to either an inadvertent autopilot activation or a pitch trim anomaly, the effects of which were compounded by aileron and/or rudder trim inputs; however, it was not possible to determine the exact nature of the initiating event. Regardless of what the initiating event was, evidence from Cessna flight test records, postaccident simulator tests, and the NTSB’s postaccident performance study indicated that the result would have been controllable if the captain had not allowed the airspeed and resulting control forces to increase while he tried to troubleshoot the problem. Evidence showed that, despite the abnormal control situation, the captain was able to maintain control of the airplane without much exertion when the airplane was operating at a relatively slow airspeed shortly after takeoff, but he increasingly struggled as the airplane accelerated and the control forces increased. By allowing the airplane’s airspeed to increase while engaging in haphazard and poorly coordinated troubleshooting efforts, the pilots allowed an abnormal situation to escalate to an emergency. Therefore, the NTSB concludes that, regardless of the initiating event, if the pilots had simply maintained a reduced airspeed while they responded to the situation, the aerodynamic forces on the airplane would not have increased significantly; at reduced airspeeds, the pilots should have been able to maintain control of the airplane long enough to either successfully troubleshoot and resolve the problem or return safely to the airport. If the accident pilots had sought guidance for either of the potential scenarios in the airplane’s checklists or manuals, they would not have found it, because no such guidance existed. However, AFM guidance for a different control anomaly (a jammed elevator trim) suggests that, in the face of an unexpected and unidentified control problem, a pilot should maintain a configuration and airspeed already proven to be controllable.83 In partial response to an NTSB safety recommendation issued after a series of upset-related air carrier accidents (A-96-120), an industry team developed an airplane upset recovery training aid designed to “increase the ability of pilots to recognize and avoid situations that can lead to airplane upsets and to improve their ability to recover control of an airplane….” The FAA has issued an NPRM, which,84 if adopted as a rule, would require minimum standards for 14 CFR Part 121 flight crewmember upset and loss of control training that references the airplane upset recovery training aid.

[ANALYSIS] The first officer could not find the circuit breakers and asked the captain where they were located. The captain did not respond to the first officer’s question; instead he instructed the first officer to declare an emergency and advise ATC of their intention to return to MKE. The first officer subsequently used an incorrect aircraft call sign in his radio transmission, and the captain corrected him. About 10 seconds before the end of the recording, the captain instructed the first officer to take control of the airplane while he, the captain, tried to pull circuit breakers. Seconds later, the captain stated, "…we're not holding it," and about 5 seconds later the recording ended. Performance data suggest that almost immediately after the first officer took control, the airplane rolled to the left and descended into the lake. It is clear that the captain had at least marginal 89 For example, when the pilots were inbound to MKE, the captain told the first officer, "Don't hesitate to ask a question…" When the first officer began to prematurely complete the before-landing checklist, the captain told him twice to wait until he talked with the control tower. Additionally, the captain coached the first officer to get the instrument landing system approach properly displayed on his flight instruments; then, during the landing rollout, the captain stated that they would forego the use of spoilers, indicating, "I should have called for 'em. I didn't want you being confused on it." As they taxied to the FBO after landing, the captain asked the first officer to call the FBO to announce the flight’s arrival; the first officer asked, "What switch do I use for com two?" The captain replied, "Number two. Turn your knob. There you go." NTSB Aircraft Accident Report 55 control of the airplane, and the airplane had successfully reversed course to return to the airport until the captain transferred control to the first officer. It is possible that the captain believed that pulling circuit breakers was the best course of action to avert a loss of control. However, the captain’s efforts to maintain control of the airplane had been successful to that point, and he could have significantly eased the ongoing and increasing control problem by reducing airspeed. The first officer provided little or no support to the captain during either flight and was arguably a distraction because the captain had to monitor the first officer’s actions as well as perform flying pilot duties. In addition, as the emergency escalated, the captain began to take over the radio calls while he continued to handle the airplane and troubleshoot the control problem. Although these distractions might have been benign during a normal flight, they were critical during the accident sequence, especially the later stages. The NTSB concludes that, as a result of the first officer’s poor flying skills, his lack of airplane systems knowledge, and both pilots’ lack of communication and coordination, the first officer provided little help to, and likely hindered, the captain during his attempts to deal with the flight control anomalies during the accident flight.

[ANALYSIS] As previously noted, the autopilot and yaw damper engage buttons are identical and are located adjacent to each other on the autopilot console, and anecdotal evidence indicates that inadvertent activation of the autopilot when activation of the yaw damper is intended is not an uncommon occurrence.75 If the first officer had inadvertently engaged the autopilot instead of the yaw damper (which is not inconsistent with his limited proficiency in the accident airplane) and the situation went unnoticed (and, therefore, uncorrected) by the captain, the autopilot would have tried to hold the heading and climb pitch attitude on which it was engaged; the associated heading was likely on or about the runway’s magnetic heading of 009°. The autopilot inputs would result in a control wheel force toward that heading whenever the pilots tried to turn the airplane in a different direction. It is likely that, at some time during the accident sequence, the first officer manipulated the aileron and/or rudder trim in an effort to help the captain with the control problems he was experiencing. The CVR recorded several trim-related comments made by the first officer.76 The last of these comments, “how’s that. Any better?” occurred about 90 seconds after takeoff and after the captain complained about the trim and could plausibly have referred to an effort he made to assist the captain. The NTSB’s airplane performance study indicated that the airplane banked sharply after this statement. The CVR then recorded a grunting noise made by the captain, and the performance study indicated that the airplane subsequently returned to its previous heading. The CVR evidence indicates that the pilots did not coordinate with each other regarding trim inputs, and it is likely that the first officer’s trim inputs aggravated, rather than ameliorated, the situation. Almost immediately, the captain told the first officer to advise ATC that they were returning to MKE. (Pilot coordination will be discussed in section 2.3.) Because the captain would normally have hand-flown the initial portion of a flight, he would not have expected the autopilot to be engaged. Additionally, because he did not request trim adjustments, the captain would not have anticipated having to counter trim-related forces while haphazardly troubleshooting the problem. As a result, the captain’s troubleshooting during the accident sequence was confused. According to the NTSB’s performance study, the lateral and directional handling problems that would result from inadvertent autopilot activation and mistrim under these circumstances would result in control forces that were light at first and larger as the airplane accelerated and turned further from the runway extended centerline heading.

[ANALYSIS] The Safety Board considered it plausible that the pilot's attention may have been directed to the movement of the left engine instruments after the right engine instruments became static. If he perceived that the left engine was the problem, he may have reacted to that perception by applying the rudder correction for a left engine failure rather than continuing with ‘he rudder deflection appropriate for a right engine failure. The attentiveness of the crew to flight instruments during the emergency anc a coordinated response to the indications were critical to maintaining control because a swept wing airplane, such as the DC-9, when in a sideslip, will tend to roll unless ecrrectiv2 action is taken. Either the sideslip must be reduced by appropriate rudder deflection or the lateral controls must be deflected to counter the rolling tendencies. The rolling tendency due to sideslip will increase as the lift on the wings increases. For a given G load (lift) and sideslip angle, a certain amount of lateral contro! deflection will be required to counter the roll. As the G load inereases, additional lateral control deflection will De required te counter the roll. in this case, elevatcr contre] input cxused the G load to increase from about 9.3 Gs to about 1.8. Gs from the 5th to the 11th second. In the presence of the sideciip angle and increasing acceleration load, the lateral control deflection would have tc be approximately doubled to maintain a constant dank angle. If the pilot established a lateral control deflection at the fourth to fifth second to compensate for the sideslip angle and then was not monitoring the roll attitude as the G load increased, the airplane would rofl further to the right. The data indicate that by the time the stickshaker came on, the airplane was in a significant roli attitude to the right and the positive G load was increasing. All of the above conditions should have deen evident to the flightcrew by reference to the flight instruments. The captain, had the individual ability and the responsibility to scan the instruments and to take corrective acticn. However, an appropriate coordinated flighterew response also would involve actions by the first officer to assist the captain in diagnosing and responding to the problem. The redundancy provided by the first officer is one of the basic tenets of cockpit resource management. 2.5.5 Cockpit Resource Management The investigation revealed that Midwest Express did not have a formal training rogram in cockpit resource management, which is also known as erew coordination. However, the Safety Board belNeves that with a low pilot to supervisor ratio. the airline could, and probably did, monitor closely the performance of its crewmembers, both as pilots and as individuals participating in the joint operation of flights.

[ANALYSIS] About 4 seconds after the right engine failed, the airplane began to yaw rapidly to the right, as indicated on the FDR deta by the 20° ~-45- heading change from the third to the seventh second (after the right engine failure), while the radar data indieated that the airplane was continuing in a relatively straight track. The yaw rate was greater than that which would have occurred due tc a sudden toss of right engine thrust, or @ sudden release of the rudder pedal force used to compensate for the asymmetrical thrust. The Safety Board determined that the sideslip angle reached about 15° and that the total yaw reached 20° during this interval. The airplane heading deviated even ferther to the right and at a more rapid rate from the eighth to the ninth second, indicating that a large roil angle was developing. As the airplane started to descend, the normal acceleration forces increased rapidly. About 1 second later, 9.6 seconds after the right engine failure, the stali warning stickshaker act.vated when the normal acceleration indication was 1.5 G. Normal acceleration increased to about 1.8 G while the descending right turn continued, indicating that the airplane entered an accelerated stall at about 148 Knots. The airplane crashed about 5 seeonds later. (See figure 5.) The large heading changes which occurred later than the ninth second after the engine failed could not have occurred without the development of a large roll angle, in addition to right rudder deflection. Also, the ground track was not consistent with heading change due to roll angles, and the low normal accelerations (less than 1 G), which were recorded in this interval, wouid diminish the effects of roll angle on the heading change rate. Therefore, the Safety Board conchided that the sudden heading change which occurred before the eighth second after the right engine failure was caused by a yawing moment, rather than a rolling moment. The Safety Board believes that the configuration of flight 105 did not change after gear retraction. The flaps probably remained at 20° deployment until impact when they were driven farther downward to about 28° In other DC-9 aecidents, the Safety Board has found similar flap movement during the impact sequence. There was no evidence that the fiighterew initiated efforts to land the airplane. The investigation revealed <hat the flightcrew was medically and operationally qualified for the flight. They had received sufficient rest, and no evidence of adverse stress-related factors was found. Weather and air traffie control were not considered to be factors in the accident. N1O0ME was certified, maintained, and equipped in accordance with applicable FAA regulations and approved procedures. The original airplane certification process had required demonstration of relevant handling qualities of the airplane, including conditions normally encountered in the event of sudden loss of thrust of either engine. The results of this investigation did not reveal any handling characteristics of the DC-9-14 which were inconsistent with the original standards for certification of the airplane.

[ANALYSIS] Any reduction in left engine thrust that occurred before stickshaker would have reduced the yawing of the airplane which occurred after the right engine failure. While possibly necessitating a foreed landing. a left engine power ioss should not have precipitated, or even contributed to, a loss of control of the airplane. However, the reduction in ieft engine power could have confused the crew. A detailed discussion of the Safety Board's analysis of the leit engine mechanical condition and operation is contained in appendix i. 25 Evaluation of Flightcrew Response The flight demonstration of a DC-9-14 airplane showed that with a sudden loss of right engine thrust at 170 knots, lateral and directional control could be maintained even if the cilot took no immediate action to deflect the rudder. Under thesc conditions, the airplane experienced about an 8° heading change and developed about 5° of sideslip within 4 seconds. About 30° of control wheel deflection, or 8° left deflection of the rudder, was required to maintain 4 wings-level attitude of the airplane. The flight demonstration was conducted with about 9,500 pounds of continuous thrust on the left engine (in-flight takeoff power). The sound spectrum examination disclosed that, on the accident airplane, the left engine thrust dropped from 19,750 pound (initially) to about 9,500 pounds after 2 to 3 seconds and to 5,500 pounds at the time of the loss of control. Because of the reduced asymmetric thrust, the yawing moment would have been reduced considerably on the accident airplane, similar to the demonstration airolene. Since there was no difficulty in compensating for the thrust asymmetry on the demonstration flight, the Safety Board concludes that the yawing moment should have been controllabie in the accident airplane. Since the airplane maintained its heading for the first 3 to 4 seconds after the rignt engine failed, it was concluced that the rudder was deflected properly to the left during that interval. However, based upon caleviations of the airplane's vawing response and resultant ground track for various rudder deflections and roll angles, the Safety Board Getermined that the large heading change and sideslip angle that developed after the first 4 seconds could not have been accomplished without a deflection of the rudder to the right, followed by a9 roll to the right 4 to 5 seconds later. Based upen the known performance of a DC~9-14, the closest duplication of the heading change which occurred on the accident flight (indicated by the FDR) would require the rudder to be deflected 6° to the left for about 3 seconds, followed by a rapid return of the rudder to neutral, then deflection of the rudder 12° to the riett about 3 seconds after the right engine failure. Returning the rudder to neutral and holding neutral rudder, after initially applying rudder to correct for differential engine thrust. wouid not have created the heading change rates which were indicated by the FDR data.

[ANALYSIS] The actions of the flying pilot, in response to the emergency, should be closely monitored by the nonflying pilot, and the nonflying pilot should be monitoring instruments in support of the flying Dilet t ensure prompt and correct response, in accordance with published emergency proce jures. Takeoff emergencies typically are critical operations because of low altitude and iow speed at their outset. Even though the initial response of the flying pilot may be reflexive, involvement by the sonflying pllict is ¢ssential to a proper crew response to the emergency. In this accident, the cockpit voice recorder reveaied the absence of emergeney eallouts from either pilot and no response from the first officer when the captain requested assistance. Because the first officer responded to Milwaukee Tower, it is clear that he was not incapacitated. Hewever, his failure to communicate with the captain and 297 Aireraft Accident Report--"Galaxy Airlines, Inc., Lockheed Electra L-188C, N5532, Reno. Nevada, January 21, 1985" (NTSB/AAR-86/01)}}. ~62- his possible misjudgment of the seriousness of the situation may have led to an uncoordinated crew respense to the emergency. The Safety Board believes that a breakdown in crew coordination was 4 significant factor in the accident. 2.8 FAA Surveillance and Oversight 2.6.1 Midwest Express Airlines The Safety Board believes that the FAA oversight of Midwest Express procedures and training during certification and ongoing day-to-day activity in the carrier's first 2 years of operation was less than optimum and probably suffered as a direct resuit of the inexperience of the POL The POI testified that she devoted only 20 pereent of her worktime tc Midwest Express, her only FAR 121 scheduled passenger airline, and that she was still obligated to perform routine general aviation duties. The Board noted that the POI had no previous FAR 121 air carrier experience, that she was not rated in a turbojet of the category and class used by the airline, and that she had not received any formal training in the DC-9 airplane used by the certificate hoider for which she was responsible. In fact, she had no turbojet pilot experience. Neither did the POI have available for consultation or assistance air carrier inspectors or DC-9 rated pilots in her own office. Although the POI used the services of air carrier inspectors assigned to other offices to ful’ i her responsibilities, it is apparent that this practice reduced her exposure to the operation of the airline. Apparently, she had become so dependent on other inspectors in surveilling Midwest Express that her own role was reduced primarily to scministrative matters. The absence of first-hand knowledge of the carrier and her lack af experience in turbojet air carrier operations severely handicapped her ability to perform the quality of surveillence required to detect shortcomings of a FAR 121 airline operation.

[ANALYSIS] S* PREDICTED NOMINAL GROUND IMPACT PC:NT “NN TAKEOFF FLIGHT P ESTIMATED POINT OF EJECTION ie) 250 FT 500 FT a a I LONGITUDINAL RUNWAY SCALE DISTANCE FROM END OF RUNWAY {{FT}} Figure 6.—-Caleulated nominal ground impact point and actual locations of engine debris adjacent to runway 19R. r Figure 7 shows that undefleeted parts are initialiy ejected in a direction approximated by the broomstick method; in this ease, 20° outbeard of vertical (away from the airplane fuselage) with high energy and speed. To have struck the Midwest Express fuselage, the parts would have had to have been deflected in excess of 65° and thus virtually all energy would have been lost. Study of the DC-9-14 control system revealed that all of its components which pass through the aft fuselage pass the engines below the cabin floor level. A were protected by multiple layers of aircraft structurc. Therefore, for ejected engine parts to nave struck and damaged any of these control system components, the ejected parts first would have to have been deflected more than 120° from their initial tangential ejection path and then would have to have penetrated and continued through engine cowling, engine pylon, fuselage, possibly fuselage supporting structure, the cabin floor, and possibly several intercostal flocr beams. Having reached the control system component(s), sufficient energy would have to have remained to disable the system components. The Safety Board believes that the possibility of ejected engine parts reaching internal control system components was extremely remote, if not impossible. The possibility of parts penetrating the fuselage at a point farther aft and damaging control components in the vertical fin would have been even more remote. Relatively low velocities would have been required for parts to have progressed in that Girection and to have struck the sirplane, while high energy would have been required to penetrcte the fuselage structure. Moreover, examination of the control system revealed redundancies which would have allowed the flightcrew to maintain full control of the airplane even if some control svstems had been disabled. Also, it was found that the rudder hydraulic actuator, which controls rudder movement by hydraulic pressure or by transferring control input to the aerodynamic tab, showed no evidence of preimpact damage. Additionally, the rudder power shutoff vaive was found with a bent control rod and discoloration, consistent with rudder hydraulic power on at ground impact.

[ANALYSIS] Returning the rudder to neutral and holding neutral rudder, after initially applying rudder to correct for differential engine thrust. wouid not have created the heading change rates which were indicated by the FDR data. Likewise, a system malfunction which would cause the rudder to trail in a near-neutral position would be inconsistent with the FDR-indicated heading change data. The demonstration flight in a DC-9-14 airplane showed that the airplane had no control characteristics which were inconsistent with the applicable certification standards; the airplane was feund to be fuy controliable in an engine-out flight environment, even without using rudder (the primary control for correeting vaw and maintaining heading) to correct for yaw. Having found no evidence or airplane performance basis for concluding that there was a control system failure or malfunction, the Safety Board concludes that the rudder deflection, which occurred beginning 4 to 5 seconds after the right engine failure, was the result of the flightcrew's improper response. Based on the analysis of the airplane performance, the yaw generated by the incorrect rudder deflection, combined with G loading, caused the airplane to enter an accelerated stall at an altitude tco low for recovery. In the seconds which preceded the accelerated stall and loss of eontrol, the airpiane was ina very dynamic situation. The increasing rate of roll, the sideslip, and the increase in acceleration load ali affected adversely the stall speed. Because of the rapidly changing attitude of the airplane, the pilots would not have been expected to know the speed at which the airplane would stall in aecelerated flight. Compared to the increase in stall speed, the 8-Knot error in indicated airspeed (due to static source error in a sideslip) would not have been significant. Further, the stickshaker stall warning system would not and did not provide the customary 4 to 5 seconds warning which is typical of that system because of the rapid entry into the stall. The Safety Board concludes that the stall occurred because the flighterew dic not diagnose the nature of the emergency correctly, applied ineorreet rudder control about 4 to 5 seconds after the right engine failure, and applied nose-up elevator control which increased the G loads. The nose-up elevator contro}} input would have been a normal response to correct for the pitech-over maneuver and the reduction in pitch attitude which was precipitated Dy the rudder pedal induced roll and was consistent with the rapid deceleration of the airplane. The rapid deceleration would have resulted in a vestibular perception of downward pitching of the nose of the airplane. The Safety Board believes that more effective scanning of the flight and engine instruments by the pilots of fight 105 would have enabled them to maintain control of the airolane and to properly evaluate the powerplant anomalies.

[ANALYSIS] The Safety Board acknowledges that the runway marking and lighting were in accordance with FAA requirements, and does not consider them to be factors in this accident, except to the extent that they may not have provided the pilot with sufficient cues to cause him to be more attentive to the controller’s clearance. Runways 30R and 30L have complex approach lighting systems, which are especially visible at night. At the time of the accident, the white runway edge lights of runway 31 were operating at a dimmer setting than those of runways 30R and 3OL, which is standard practice at STL. The Safety Board believes that the dimmer lights on runway 31 were not sufficient to distract the pilot from his preconception that runway 30R was his intended departure runway. Finally, as the Cessna 441 pilot proceeded from taxiway Romeo into position on runway 30R, he entered the runway at an intersection 2,500 feet from the threshold. According to the AIM, an intersection clearance can be requested by the pilot or initiated by the controller. The Cessna pilot did not request an intersection takeoff, nor did the ground controller indicate that the pilot should expect an intersection departure, and the pilot should not have entered the runway at an intersection, without specific clearance to do so. While the Cessna 441 pilot’s entry onto the runway at an intersection should have been a final cue that his notion of being cleared to runway 30R was incorrect, the cue was apparently not sufficient to cause him to question his perception that he had been cleared to runway 30R. 2.2.3 Communications Effective radio communications between the Cessna 441 pilot and ATC were critical to establishing a mutual understanding of intentions. The ground controller’s multiple frequencies were congested with almost continuous communication, which resulted in several simultaneous transmissions in the 20 minutes before the accident. Additionally, there was some indication that the Cessna 441 pilot might have experienced communication radio difficulty. Specifically, the pilot complained about his communication radios during the inbound flight to STL, and several subsequent transmissions were garbled. Under these circumstances, it was especially critical for the pilot to ensure that effective communications were taking place. The Safety Board noted that the Cessna pilot did not sta!e the departure runway in any of his clearance readbacks. Although critical item readbacks have always been considered important in airborne operations, until recently, there was no requirement for critical item clearance readbacks for surface operz:ons. This omission was addressed in Safety 33 Recommendations A-95-33 and -34, which were issued, as mentioned in Section 1.18.5 of this report, during this investigation. In response to the recommendations, the FAA stated that it anticipates changing the AIM and Advisory Circulars 61-2 1 A, “Flight Training Handbook,” and 21-23B, “Pilots Handbook of Aeronautical Knowledge,” urging pilots to read back in full their runway assignment when operating at airports with more than one runway.

[ANALYSIS] During the on-scene investigation, several local pilots acknowledged that the proximity of runway 31 to the Midcoast ramp created a situation where pilots could inadvertently enter onto runway 3 1 without recognizing that they were on a runway. The Cessna 441 pilot had an airport diagram available in the cockpit, and may have referred to it during his outbound taxi. However, it is also possible that he believed the taxi route to runway 30R was obvious and thus did not pay much attention to the diagram. Even if he had referred to the airport diagram, he may have had difficulty discerning runway 31 on it, due to dim lighting in the cockpit, and the competing tasks that included taxiing the airplane and performing checklists. While runway 3 1 had markings, signage, and lighting consistent with FAA airport certification requirements, had several elements been slightly different, they might have triggered the pilot to question his notion that runway 30R was the assigned departure runway. These elements include runway width, displaced threshold, and runway markings and lighting. Runway 31 is 75 feet wide, which is typical of taxiways at STL. In contrast, runways 30R and 30L are 150 and 200 feet wide, respectively. At the time of the accident, a 1,838-foot displaced threshold was incorporated into the runway 3 1 marking/lighting scheme. The markings on the approximately 800-foot-long portion of runway 3 1 on which the Cessna pilot back-taxied consisted of a series of white arrows pointing toward the numbers. The runway 3 1 numbers were located at the end of the displaced threshold, near the intersection of runway 3 1 and taxiway November. The Cessna pilot’s taxi route did not go past the numbers. I 32 Along the displaced threshold, the runway lights had split red and white lenses, situated so that the white side of the lens was presented to the Cessna pilot as he back-taxied. This would have been a clue to the pilot that he was on a runway. However, the red side of the lens would have been visible to an airplane on approach for the runway, or to a pilot holding in position on runway 31 for departure. Because of the displaced threshold marking scheme, the Cessna pilot could not have seen the numbers for runway 3 1. Had he seen the numbers, the pilot might have been cued to question the controller as to the controller’s intentions. The Safety Board acknowledges that the runway marking and lighting were in accordance with FAA requirements, and does not consider them to be factors in this accident, except to the extent that they may not have provided the pilot with sufficient cues to cause him to be more attentive to the controller’s clearance. Runways 30R and 30L have complex approach lighting systems, which are especially visible at night.

[CONCLUSIONS > FINDINGS] There was no mention of the occasional use of runway 31. 10. The controller clearly referenced runway 3 1 in two separate transmissions. In both cases, the pilot acknowledged the clearance, but did not read back the runway assignment. Had the controller used more precise phraseology in the issuance of the initial taxi clearance, the Cessna 441 pilot may have noted the proper departure runway. 11. Had the Cessna 441 pilot volunteered, or had the controller requested, 48 confirmation of the assigned runway, the pilot’s error may have been detected and the accident prevented. 12. Air traffic control personnel were not able to maintain visual contact with the Cessna 441 after it taxied from the well-lighted ramp area into the runway/taxiway environment of the northeast portion of the STL airport. 13. An operational ASDE-3, particularly ASDE-3 enhanced with AMASS, could be used to supplement visual scan of the northeast portion of the STL airport surface. 14. The MD-82 flightcrew stated that they did not observe any external lights on the Cessna 441 before impact. When the Cessna 441 was in position for departure on runway 30R, the most conspicuous exterior lighting was directed forward, and, with the possible exception of wing anticollision/strobe lights, would not have been visible to the MD-82 flightcrew. 15. It is likely that the wing anticollision/strobe lights were not operating when the collision occurred. 16. Pilot training for surface movement can be improved in both air carrier and general aviation areas. 3.2 Pm bable Cause The National Transportation Safety Board determines that the probable cause of this accident was: the Cessna 441 pilot’s mistaken belief that his assigned departure runway was runway 30R, which resulted in his undetected entrance onto runway 30R, which was being used by the MD-82 for its departure. Contributing to the accident was the lack of ATIS and other ATC information regarding the occasional use of runway 31 for departure. The utilization of an operational ASDE-3, and particularly ASDE-3 enhanced with AMASS, could have prevented this accident. 49

[ANALYSIS] She also indicated that it was the pilot’s habit to taxi with the airport diagram in front of him. The current STL airport diagram approach chart was located in the cockpit area of the Cessna 441 wreckage. Additionally, the pilot’s flight logbook revealed that he had landed once before at STL. The previous flight was a daytime operation and occurred approximately 10 months before the accident. 2.2.2 Alternate Scenario: Pilot’s Pleconceplion that Runway 30R was his Ihparture Runway The evidence indicates that it was unlikely that the pilot was lost, but rather that he had a preconception that he would be departing on runway 30R and thus did not register the ground controlle.-‘s clearance to runway 3 1. Several situational cues may have reinforcec ‘ie Cessna 441 pilot s preconception that runway 30R was his assigned departure runway. ? he Cessna 441 pilot had landed on runway 30R about 18 minutes before he received the taxi 31 clearance to runway 3 1 for his departure. The “quick turnaround” nature of the flight may have added to the Cessna pilot’s belief that he would be departing on runway 30R. Also, from the time he approached STL for landing until he taxied out for takeoff, all traffic had landed and departed on runways 30R and 30L. The ATIS that was current during the time the pilot operated in the STL area listed runways 30R and 30L as the active runways for arrivals and departures at STL. The STL controllers did not typically list runway 31 as an active runway on the ATIS, as runway 31 was only occasionally used as a departure-only runway. Also, the STL controllers did not typically treat runway 3 1 as if it were an active runway; for example, when the Cessna 44 1 pilot cleared runway 30R on his inbound flight, his taxi clearance to the Midcoast ramp did not include a clearance to cross runway 31. The Safety Board believes that if runway 31 had been referenced as a runway for occasional general aviation departures on the ATIS broadcast, the pilot may have been more attentive to the controller’s taxi clearance and runway assignment. Another situational cue that could have reinforced the Cessna 441 pilot’s notion that runway 30R was his departure runway was the fact that when he began to taxi outbound from the Midcoast ramp on taxiway Whiskey (150 feet long), he almost immediately encountered runway 3 1, unlike the more typical airport layout in which a ramp exit leads to a parallel taxiway en route to the runway. During the on-scene investigation, several local pilots acknowledged that the proximity of runway 31 to the Midcoast ramp created a situation where pilots could inadvertently enter onto runway 3 1 without recognizing that they were on a runway.

[ANALYSIS] Had the equipment been operational, it would have displayed the Cessna 441’s position at the intersection of taxiway Romeo and runway 30R for about 3 minutes before the collision. 3 30 The circumstances of this accident indicate that the pilot of the Cessna 441 unintentionally deviated from the taxi clearance he received from the ground controller and taxied onto the active runway being used by the MD-82. The Safety Board’s investigation examined possible reasons why the Cessna pilot might have believed that runway 30R was his departure runway. 2.2 Cessna 441 Pilot Performance The Safety Board believes that several personal factors may have contributed to the Cessna 441 pilot’s deviation from ATC instructions. According to the pilot’s wife, the accident occurred at a time of night when the pilot normally went to sleep, and he may have been tired. Company personnel reported that such late trips were unusual. Although the pilot’s workrest cycle is not consistent with chronic sleep loss, the fact that he was operating during a period in which he was normally at rest may have had some effect on his performance and level of attentiveness. Additionally, the pilot of the Cessna 441 had commented that it was going to snow in Iron Mountain that night. Midcoast personnel stated that the pilot seemed anxious to go home, a behavior that they considered normal among pilots at that time of night. The combination of the time of day and his desire to return home before the weather deteriorated may have contributed to the mistaken actions of the Cessna 441 pilot, who was generally described in positive terms for his cautious and safe attitude. 2.2.1 Scenario: Pilot Became Lost During Airport Ground Operations The Safety Board considered the possibility that the pilot intended to take off from runway 3 1, as directed, but became lost on the airport and ended up in position to take off on the wrong runway. However, the pilot did not indicate confusion in his radio responses to the taxi clearance, and radar data indicated no hesitation in his taxi route. (See Appendix E.) The passenger who chartered the Cessna 441 to STL before the accident flight reported that on a previous flight when the pilot of the Cessna 441 had become uncertain of his position at an airport, he had stopped taxiing until he determined his location. She also indicated that it was the pilot’s habit to taxi with the airport diagram in front of him. The current STL airport diagram approach chart was located in the cockpit area of the Cessna 441 wreckage. Additionally, the pilot’s flight logbook revealed that he had landed once before at STL.

[ANALYSIS] 2. ANALYSIS 2.1 General The MD-82 flight and cabin crews were certificated, trained, and qualified for their duties. No physiological factors or unusual cockpit distractions existed that would have precluded the flightcrew from seeing the Cessna 441 on the runway. However, environmental factors, such as the darkness and poor conspicuity of the Cessna 441, contributed to the flightcrew’s failure to see the Cessna 441 on runway 30R. The pilot-in-command of the Cessna 441 was certificated, trained, and qualified for the charter flight. The local controller and ground controller were certificated, trained, and qualified for their duties. No physiological disabilities were apparent that would have detracted from their ability to perform at an acceptable level on the evening of the accident. However, the darkness, poor conspicuity of the Cessna 441, and physical distance between the ATC tower cab and the northeast portion of the STL airport contributed to the controller’s failure to see the Cessna 441 on runway 30R. STL airport markings, signs, and lighting near and along the taxi route of the Cessna 441 conformed to FAA standards. Both airplanes were maintained in accordance with the applicable directives, and there was no evidence of any deficiency or malfunction that would have contributed to the collision. Weather conditions were well above the criteria for VFR. In postaccident interviews, neither the flightcrew of the MD-82 nor the air traffic controllers identified environmental factors, other than darkness, as a constraint to the normal performance of their duties. The air traffic volume in the St. Louis area during the time of the accident was moderate, and traffic complexity was routine. However, the ground controller was working seven frequencies at the time of the accident, and several instances of simultaneous transmissions occurred in the 1 l/2 hours before the accident. The ASDE-3 airport surface detection equipment was installed at STL, but had not yet been commissioned and was not operational on the night of the accident. Although the equipment had not yet been commissioned, it would have been available to controllers except that there had been a computer hard drive failure. Had the equipment been operational, it would have displayed the Cessna 441’s position at the intersection of taxiway Romeo and runway 30R for about 3 minutes before the collision. 3 30 The circumstances of this accident indicate that the pilot of the Cessna 441 unintentionally deviated from the taxi clearance he received from the ground controller and taxied onto the active runway being used by the MD-82.

[ANALYSIS] The FAA should require under 14 CFR Part 139 that airport operators place appropriate runway or taxiway signs at cach intersection along alrport taxiways to designate either the intersecting taxiway or runway. The crew of KAL 084 did not Indicste in their statements that they saw the fully illuminated sign designating runway 6L/24R. Several factors may have contributed to the failure of the crew of KAL 084 to notice this sign, even though it was fully {{iluminated. The sign was dirty, which reduced the contrast between its background and lettering. Since the airport surfaces were obscured partially by snow, frost, and ice, the crew was looking intently for ground markings. Moreover, the Visibility was restricted, which further Hmited the crew's ability to see the sign, particularly since the location of the DC-10 cockpit is about 30 feat above the ground increases the slant range from cocknit to guidance signs placed aside taxiways and runways. ~-23- Contributing to the crew's failure to notice the runway sign was that, despite the different purposes that the runway and taxiway signs serve, the signs had common shape, color, and dimensional characteristics, The runway and taxiway signs had identical amber backgrounds with black lettering. The characters on the signs were identically sized. The signs, which were the same height, differeg only in their width according to the number of characters on the sign. The Safety Board is concerned that in similar situations other flighterews or vehicle operators could inadvertently enter an active runway. Runway and taxiway intersection signs should reflect, in their sizes, shapes, colors, and dimensions, the particular route they mari; a sign identifying a taxiway intersection should have a different appearance from a sign identifying a runway, and these signs should then de installed at airports certificated under 14 CFR Part 139. 2.7 Runway incursions The October 1978 ASRS article concerning human factors associated with runway incursions, as well as the three subsequent acciderits described earlier, substantiates problems and causal elements similar to those in this accident. While the December 19, 1983, accident at Anchorare and the collision at Sioux. Falls involved air traffic contro}}, the accident at: Madrid was similar to this accident. The Aviaco Airlines DC-8 pilot did not taxi as instructed at Barajas Airport during restricted visibility conditions. While the KAL 084 crewmembers did not ignore tower instructions, the factors of crewmember disorientation, cockpit coordination, and pilot technique cited in the ASRS article were evident in this accident. Flighterews must be especially vigilant during taxi, hold, and takeoff operations and must make extraordinary efforts if needed to stay aware of their position on the airport at all times. Crew coordination procedures should be enhanced and particular alertness should be practiced when visibility is reduced by inclement weather. I }} 3. CONCLUSIONS

[FINDINGS] There was no taxiway guidance sign at the intersection of taxiway W-1 and the east-west taxiway. Operators of airports certificated under 14 CFR Part 139 are not required to place standardized signs at each taxiway/runway and taxiway intersection. Runway signs should be sufficiently different in design from taxiway signs so thet they alert the operators of ail surface vehicles and airplanes of the nature of the intersection. 17. Lighted runway/taxiway signs should be inspected daily to ensure their operability and maintained as required. 3.2 Probable Cause The National Transportation Safety Board determines that the probable causes of ths accident were the failure of the pilot of Korean Air Lines Flight 084 to follow accepted procedures during taxi, which caused him to become disoriented while selacting the runway; the failure of the pilot to use the compass to confirin his position; and the aceision of the pilot to take off when he was unsure that the aireraft was positioned on the correct runway. Contributing to the accident was the fog, which reduced visibility to a point that the pilot could not ascertain his position visually and the control tower personnel could not assist the pilet. Also contributing to the accident was a lack of gible taxiway and runway signs at several intersections passed by Flight 084 while it was a oe SNH A ROPE 9 “ SORRY BENE LE AER se SEES gy I ERRATA ET Rp cor Fete ae: eee seer fe vad . aed ‘ f > 4. RECOMMENDATIONS I As 4 result of this accident investigation, the National Transportation Safety Board recommended that the Federal Aviation Administration: Require that airports certificated for air carrier operations install signs at all runway and taxiway entrances, exits, and intersections that indicate the identity of the runway or taxiway. (Class ii, Priority Action) (A-84-98) Require that the graphics on taxi:vay/runway identification signa be standardized and of sufficient size to enable them to be legible to aircraft crewmembers in all meteorological conditions in which air carrier operations are authorized. (Class i, Priority Action) (A~-84--99) Require that airport operators inspect and tmatntain the lights illuminating airport taxiway/runway identification sigs as part of the cay rita inupection requizements. (Cless I, Friority Action) A-84-100 Require at all airports certificated for air carrier operations that uniform signs be installed which are classified by function (e.g., runway entrance, runway exit, taxiway intersection) with each function having a unique shape, color, end/or size so that runway entrance signs are distinguishaole from all other advisory siens on airport property. (Class Ii, Priority Action) (A~84-101) Require that air carriers incorporate in training of their crewmembers procecures and responsibilities during ground operations in restricted visibility conditions, to enable them to operate sefely in such conditions. (Class I, Priority Action) (A-84-102) - BY THE NATIONAL TRANSPORTATION SAFETY BOARD — /s/ JIM BURNETT Chairman /s/ PATRICIA A, GOLDMAN Vice Chairman /s/ GH, PATRICK BURSLEY Member VERNON L, GROSE Member August 9, 1984 oh emedeal Fey OSU Se ne ara ed ie sere Preceding page blank = -27- APPENDIXES APPENDIX A INVESTIGATION AND HEARING

[FINDINGS] Flighterews must be especially vigilant during taxi, hold, and takeoff operations and must make extraordinary efforts if needed to stay aware of their position on the airport at all times. Crew coordination procedures should be enhanced and particular alertness should be practiced when visibility is reduced by inclement weather. I }} 3. CONCLUSIONS Findings 1. Both airplanes were certificated and maintained in accordance with approved procedures. 2. There was no evidence of preaccident failure or malfunction of either airplane's structures, systems, powerplants, or flight control systems. 3. The pilot of SouthCentral Air Flight 59 (SCA 59) was properly certificated and qualified for this scheduled domestic passenger flight. His actions did not contribute to the accident. The flighterew of Korean Air Lines Flight 084 (KAL 084) were properly certificated and qualified for this scheduled cargo flight. The flighterew of both eirplanes irvolved held current medical certificates. Both the captain and the first officer of KAL 084 had extensive experience operating into and out of the Anchorage International Airport, which should have reduced the ; robability of crew disorientation while taxling in the low-visibility conditions. The decision of KAL 084's captain to use runway 32 for departure was not in accordance with KAL operating specifications. -~34~ The obscuration of runway and taxiway markings eat the airport adversely affected the performance of the flighterew of KAL 084 by causing them to give disproportionate attention to locating the runway markings. The most likely taxi route, taken in error, by KAU 684 was south along the west side of the north apron, right ontv taxiway W~-1, and right again onto runway 24k. The flighterew of KAL 084 could have determined that thsir airplane was lined up on the wrong runway if they had cross-checked their heading indicators. Based on the estimated takeoff gross weight of KAL 084, the runway length required for takeoff was 8,150 feet. Since the actual length avaliable to KAL 084 on runway 24R was about 2,400 feet, an accident would have resulted even if KAL 084 had not collided with SCA 59. . By raising the nose of his airplane and turning his sirplane slightly to the left, the captain of KAL 084 avoided inflicting extensive damage to the fuselage of SCA59 and probable fatal injuries to the crews and passengers onboard both airplanes as a result of the collision. Of the four runway and taxiway signs KAL 084 would have passed on the most likely taxi route it erroneously took, one had no illumination, one was oniy partially illuminated, and two were fully illuminated. There was no taxiway guidance sign at the intersection of taxiway W-1 and the east-west taxiway. Operators of airports certificated under 14 CFR Part 139 are not required to place standardized signs at each taxiway/runway and taxiway intersection.

[ANALYSIS] It may be chat verification of runway heading is such a rudimentary procedure that the air ecezrier believed that specialized training was not necessary. While such a belief may have been reasonable and reflective of accepted practice, that this crew failed to carry cut this basic step indicates that a deficiency which nedds. to be addressed may exist in al cerrier crew training’ and certification procedures. : I : The Safety Board cannot explain why the captein of KAL 084 devided to take off in the fase of his uncertainty as to whether his airplane was holding at runway 32, The captain stated: I I : ~21- -.+T felt unsure that the aircraft was on the correct runway....I discussed this with my copilot [the first officer}} who felt sure that we were on the correct runway. After 3-4 minutes of discussion, I considered teking runway 6R because of my uncertainty. However, the runway size and lighting appeared to be correct so I decided to take off. This statement indicates that the captain failed to recognize that his familiarity with the airport would not compensate for the limitations in other sourees of information he would use ordinarily to confirm the aireraft's location. The captain failed to exercise proper decisionmaking responsibility by relyings too heavily on the first officer's belief that the airplane was on the correct runway. Proper command procedures should hive dictated to the captain not to commence takeoff without confirming that he was holding at runway 32. The captain's statement indicates that he felt that the first officer, who had a higher level of recent experience at the airport than the captein, was more certain about the aircraft's location than the captain was. The first officer stated that, "In spite of poor visibility, our aircraft advanced and was able to get onto the east~west taxiway." The evidence indicates that KAI 084 was never on the east-west taxiway. Unlike the captain, the first officer in his statement did not manifest ary uncertainty about the nircraft's location. The Safety Board believes that the first officer's strorg belief about their location may have influenced the captain's decision to commence takeoff. The first officer's confidence regarding being on the correct runway in the face of the csptain's uncertainties constituted a slight role reversal in that the captain's overall command authority when deciding te take off was influenced by the first officer's comments. In the past, the Safety Board has encouraged assertiveness training for first officers, to exercise their responsibilities as part of the cockpit team; however, a companion responsibility for captains to exercise positive cockpit crew management must exist. In this instance, the crew concept broke down. This breakdown may have been due to the crew's intense eoncentration on the airport surface markings and runway and taxiway signs in order to confirm their location.

[ANALYSIS] All of the involved flighterew members held current medical certificates. 2.2 Weather (art ee mate fergie ede The surface visibility at Ancharage International Airport was restricted, as evidenced by the 1350 surface observation which reported 1/8 mile visibility and the 1415 cbservation which reported 1/16 mile visibility. The local controller advised SCA 59 that the RVR was 1,000 feet at 1344:18, and the RVR did not improve to 1,800 feet until 1405:28, at which time SCA 59 was cleared to taxi into position and to hold on runway 6L. An RVR of 1,800 feet was the minimum takeoff visibility for the pilot of SCA 59. The captain of KAL 084 stated that, after he began taxiing from the parking ramp, he could see the yellow taxi lines "very dimly through the heavy ice fog." He devecribed the visibility as "so poor thay it was difficult to see the taxiway markings." After the accident, the first officer of KAL 084 concluded a written statement as fellows! "It seems that I lost my sense of direction due to the heavy ice fog, and I confuse’ the east-west taxiway with the north-south taxiway.” The testricted visibility caused the flighterew of KAL 084 to experience difficulties while operating on the taxiways and runways at Anchorage international Airport and adversely affected their operational performance. 2.3 Collision Analyals According to applicable performance charts, based on the estimated TOGW of $02,790 pounds, the temperature of 15 degrees F., and the field elevation of 144 feet, the departure runway length required for KAL 084 was 8,150 feet. The distance from the intersection of runway 6L/24R and texiway W-1, where KAL 034 began its takeoff roll, 10 the departure end of runway 248 is 2,400 feet. Based on these data, it can be concluded that the atiompted takeoff by the KAL 084 flightcrew would not have been successtul even i! their takeoff run had not been interrupted by the collision with SCA 59. _ -KAL 084 was equipped with three main gea.s, one being a centered body gear, — Given the dimensions of both airplanes, and the impact marks on SCA 59, it appears that the nore gear of KAL 084 struck SCA 59 on the right windscreen at the top and grazed the skin of the right fuselage over the cockpit, missed the remainder of the fuselage, and struck the vertical stabilizer. As the captain of KAL 064 turned left to miss SCA 59, the main body gear swung to the right and struck the left wing of SCA 59, knocking the wing off outboard of the engine nacelle.

[ANALYSIS] I The primary sources of information that are ordinarily available to crewmembers for guicance on airport surfaces were either partially or completely unavailable to the crew of KAL 084. At nighttime or under limited visibility conditions, crewmembers rely on runway surface markings such as taxiway lines and runway numbers, taxiway and runway lights, and runway and taxiway signs to provide them with information concerning their location on the airport. If the visibility is adequate, or If the airport is equipped with ASDE, ground controllers can assist the aircraft crewmembers by providing information on their location. The flighterew of KAL 084 operated essentially without external information to assist them while taxiing since the visibility was restricted and the airport did not have ASDE. a 2.5 _ KAL 084 Fightcrew Medical and Behavicral Factors The meiical examination of the KAI, 084 crewmembers immediately after the accident and the toxicological testing of bicod samples did not reveal any physiological condition which might have affected their performance. Each crewmember was well rested before the flight, having been «ff ducy for over 26 hours prior to the scheduled departure time. The crewmembers were housed in facilities operated by Korean Alr Lines for employees laying over {{n Anchorage to insure that crewmembers rest in an undisturbed environment with: Korean food and a familiar atmosphere. The performance of the crewmembers cannot be attributed to fatigue resulting from excessive duty time or te Stress created by unfarmilin’ surroundings. Similarly, interviews with the crew and their statements did not revwal any significant event in their lives thet may have caused them stress or tension or affatted their decisionmaking abilities. The flight wes not significantly delayed, nor was the crew facing an imminent deadline for completing the — flight, such ae deteriorating weather at destination, curfews, or excessive duty time. — From the response of the captain of KAI. 084 to questioning, the Safety Board could not determine why a: axperienced crew, such as this crew, did not werify whether they were on the correct runway by checking their heading instruments. The Safety Board could not find any factcr which may have adversely affected the crew's vision, coordination, or decisionmaling capabilities to determine that their heading was 80° from the correst runway bearing. ‘The failure of the crew to verity the runway heading may — indicate that the initial or rocurrant training the crew received or the op?rating procedures established for KAI, crewmembers are deficient. It may be chat verification of runway heading is such a rudimentary procedure that the air ecezrier believed that specialized training was not necessary.

[ANALYSIS] The Safety Board examined several of the runway and taxiway signs at the airport to determine if all of the avaiiabie sources of ground location information external to the airplane were adequately presented to the KAL 084 crew. The KAL airplane passed four signs identifying “unways and taxiways along the route that the Board believes it took while taxiing. One of the four signs, the sign designating taxiway N-1, was not equipped for electrical illumination. At night In restricted visibility conditions when additional guidance is most needed, such as existed at the time of this crash, this sign would provide no information or guidance to flighterews, Another of the four signs was only partially illuminated, because only three of its seven lights were operating at the time of the accident. The other two signs, which identified runway 14 and runway OL/24R, were iiluminated. Airports certificated under 14 CFR Part 139 are not required to have taxiway/runway guidance signs installed. However, if the signs are instalied, 14 CFR 139.47(b) requires that. the operator "must show that any guidance signs installed at the airport are in operable condition." Por each airport certificated under 14 CFR Part 139, the FAA approves an Airport Operations Manua! (AOM), which, in part, lists key elements of the airport, such as runway lights, that are required to be inspected daily to ensure that they are in operable condition. For many airports, including Anchorage International, the approved AOM does not. include guidance signs in the list of key elements. Therefore, although 14 CFR 139.47(o) requires that the signs be in operable condition, the FAA has not euppiied guidance to the airport operators on how or when this requirement will be met. The Safety Board believes that as KAL 084 taxied along texiway W-1, the crew thought that they were on the east-west taxiway, and that when they crossed the east-west taxiway, they thought it was the north-south taxiway and continue~ to what they believed was runway 32 but was instead runway 24R. There were no Sig..3 along this ground path to indieate, first, that the taxiway they had entered was W~1 and, second, that the first intersection they then crossed was the east-west taxiway. The crew of KAI, ignate either the taxiway they were on or axied to the intersection, of taxiway W~-1 signs have been installed at both intersections to te the intersecting taxiways. The FAA should require under 14 CFR Part 139 that airport operators place appropriate runway or taxiway signs at cach intersection along alrport taxiways to designate either the intersecting taxiway or runway. The crew of KAL 084 did not Indicste in their statements that they saw the fully illuminated sign designating runway 6L/24R.

[ANALYSIS] Analysis National Transportation Safety Board A I R C R A F T Accident Report 61 the flight crew had not correctly interpreted these cues or noticed them until after it was too late to successfully abort the takeoff. The Safety Board concludes that the flight crew recognized that something was wrong with the takeoff beyond the point from which the airplane could be stopped on the remaining available runway. Runway Incursions 2.2.1.4  The FAA currently defines a runway incursion as “any occurrence in the airport runway environment involving an aircraft, vehicle, person, or object on the ground that creates a collision hazard or results in a loss of required separation with an aircraft taking off, intending to take off, landing, or intending to land.” ICAO defines a runway incursion as “any occurrence at an aerodrome involving the incorrect presence of an aircraft vehicle or person on the protected area of a surface designated for the landing and take-off of aircraft.” According to these definitions, ICAO would classify this accident as a runway incursion, but the FAA would not consider this accident to be a runway incursion because no other airplane impeded the accident airplane’s ability to take off. The Safety Board notes that the presence of another airplane on a runway should not be a consideration in determining whether a runway incursion has occurred; rather, criteria for making this determination should consider, among other things, whether an airplane’s movement is consistent with the clearances provided to the flight crew. As a result, the Safety Board concludes that, because the accident airplane had taxied onto and taken off from runway 26 without a clearance to do so, this accident was a runway incursion.155 At the Safety Board’s runway incursion forum in March 2007, the FAA announced that it planned to revise its definition of a runway incursion to align with ICAO’s definition by the end of fiscal year 2007 and that it would begin reporting runway incursions according to the revised definition in fiscal year 2008. The Board is encouraged by the FAA’s plan to adopt and apply the ICAO definition because wrong runway takeoffs should be reflected in runway incursion statistics so that runway safety trends can be accurately monitored and appropriate countermeasures can be taken. Pilot Human Factors 2.2.2  The flight crew’s performance on the day of the accident seemed to be uncharacteristic with past reports. The captain and the first officer were described favorably by company personnel, and pilots who had flown with them described both as competent pilots who had not previously demonstrated difficulty with airport surface operations. The captain was described as someone who managed the cockpit well, adhered to standard operating procedures, and demonstrated good CRM. The first officer was preparing for an opportunity to upgrade to captain and was described as someone who would have made a good captain because of his adherence to standard operating procedures. 155  Even though the FAA does not consider the SEA and MIA incidents (discussed in section 1.18.4) to be runway incursions, the Safety Board does consider these incidents to be runway incursions.

[ANALYSIS] However, as previously stated, the Board believes that these events occurred because the flight crew did not use the available cues on the airport surface during taxi and did not cross-check and confirm the airplane’s position on the runway before departure. Also, the flight crewmembers engaged in a nonpertinent conversation during a critical phase of flight (taxi operations), which caused them to lose positional awareness. In addition, before the airplane arrived at the dark runway, the first officer briefed the outage of the runway end identifier lights, he recounted that lights were out during his arrival at LEX early on the morning that preceded the accident flight, and the flight crew most likely read the NOTAM in the flight release paperwork that indicated that the runway 4/22 centerline lights were out of service. This accident is not the first one involving a wrong runway takeoff in which pilots did not use existing cues to identify the airplane’s location on the airport surface or cross-check and verify the airplane’s position before takeoff. For example, the final report on the October 31, 2000, Singapore Airlines flight 006 accident (see section 1.18.4) stated that the pilots did not verify the airplane’s position on the taxi route as the airplane was turning onto the wrong runway and concluded that the flight crew lost situational awareness and took off from the wrong runway despite numerous available cues that provided information about the airplane’s position on the airport. However, the final report for that accident cited a possible reason for the flight crew’s actions: the pilots did not adequately review the taxi route to ensure that they understood that the route to the correct departure runway required passing a parallel runway that was under construction and was open only for taxi operations. Numerous other wrong runway events have occurred. The ASRS database showed 114 reports of incidents from March 1988 to September 2005 in which flight crews of turbojet airplanes lined up on the wrong runway for takeoff. The ASRS reports indicated that the pilots involved in these events, pilots of other aircraft in the area at the time, or air traffic controllers detected the mistake either before or after takeoff. Also, on October 30, 2006, an Alaska Airlines 737 departed from the wrong runway at SEA. According to postincident interviews, the controller instructed the flight crew to taxi the airplane into position and hold on runway 34C, but the captain assumed that the airplane would be taking off from runway 34R. After the airplane departed uneventfully from runway 34R, the controller informed the flight crew that the airplane had departed from the wrong runway. Most recently, on April 18, 2007, a United Airlines Airbus A320 taxied onto runway 27, which was closed, instead of runway 30, the assigned departure runway, at MIA and began the takeoff roll. A NOTAM, which was included in the flight release paperwork, and the ATIS information broadcast, which the flight crew received, indicated Analysis National Transportation Safety Board A I R C R A F T Accident Report 76 that runway 27 was closed.

[FINDINGS] 103 Conclusions 3. Findings 3.1  The captain and the first officer were properly certificated and qualified under Federal 1. regulations. There was no evidence of any medical or behavioral conditions that might have adversely affected their performance during the accident flight. Before reporting for the accident flight, the flight crewmembers had rest periods that were longer than those required by Federal regulations and company policy. The accident airplane was properly certified, equipped, and maintained in accordance 2. with Federal regulations. The recovered components showed no evidence of any structural, engine, or system failures. Weather was not a factor in this accident. No restrictions to visibility occurred during 3. the airplane’s taxi to the runway and the attempted takeoff. The taxi and the attempted takeoff occurred about 1 hour before sunrise during night visual meteorological conditions and with no illumination from the moon. The captain and the first officer believed that the airplane was on runway 22 when 4. they taxied onto runway 26 and initiated the takeoff roll. The flight crew recognized that something was wrong with the takeoff beyond the 5. point from which the airplane could be stopped on the remaining available runway. Because the accident airplane had taxied onto and taken off from runway 26 without 6. a clearance to do so, this accident was a runway incursion. Adequate cues existed on the airport surface and available resources were present in 7. the cockpit to allow the flight crew to successfully navigate from the air carrier ramp to the runway 22 threshold. The flight crewmembers’ nonpertinent conversation during the taxi, which was not in 8. compliance with Federal regulations and company policy, likely contributed to their loss of positional awareness. The flight crewmembers failed to recognize that they were initiating a takeoff on the 9. wrong runway because they did not cross-check and confirm the airplane’s position on the runway before takeoff and they were likely influenced by confirmation bias. Even though the flight crewmembers made some errors during their preflight activities 10. and the taxi to the runway, there was insufficient evidence to determine whether fatigue affected their performance. Conclusions National Transportation Safety Board A I R C R A F T Accident Report 104 The flight crew’s noncompliance with standard operating procedures, including the 11. captain’s abbreviated taxi briefing and both pilots’ nonpertinent conversation, most likely created an atmosphere in the cockpit that enabled the crew’s errors. The controller did not notice that the flight crew had stopped the airplane short of the 12. wrong runway because he did not anticipate any problems with the airplane’s taxi to the correct runway and thus was paying more attention to his radar responsibilities than his tower responsibilities. The controller did not detect the flight crew’s attempt to take off on the wrong runway 13. because, instead of monitoring the airplane’s departure, he performed a lower-priority administrative task that could have waited until he transferred responsibility for the airplane to the next air traffic control facility.

[CONCLUSIONS] Because of an ongoing construction project at Blue Grass Airport, the taxiway identifiers 25. represented in the airport chart available to the flight crew were inaccurate, and the information contained in a local notice to airmen about the closure of taxiway A was not made available to the crew via automatic terminal information service broadcast or the flight release paperwork. The controller’s failure to ensure that the flight crew was aware of the altered taxiway A 26. configuration was likely not a factor in the crew’s inability to navigate to the correct runway. Because the information in the local notice to airmen (NOTAM) about the altered 27. taxiway A configuration was not needed for the pilots’ wayfinding task, the absence of the local NOTAM from the flight release paperwork was not a factor in this accident. The presence of the extended taxiway centerline to taxiway A north of runway 8/26 28. was not a factor in this accident. Probable Cause 3.2  The National Transportation Safety Board determines that the probable cause of this accident was the flight crewmembers’ failure to use available cues and aids to identify the airplane’s location on the airport surface during taxi and their failure to cross‑check and verify that the airplane was on the correct runway before takeoff. Contributing to the accident were the flight crew’s nonpertinent conversation during taxi, which resulted in a loss of positional awareness, and the Federal Aviation Administration’s failure to require that all runway crossings be authorized only by specific air traffic control clearances. National Transportation Safety Board A I R C R A F T Accident Report 106 Recommendations 4. New Recommendations 4.1  As a result of the investigation of this accident, the National Transportation Safety Board makes the following recommendations: —To the Federal Aviation Administration: Require that all 14 Code of Federal Regulations Part 91K, 121, and 135 operators establish procedures requiring all crewmembers on the flight deck to positively confirm and cross-check the airplane’s location at the assigned departure runway before crossing the hold short line for takeoff. This required guidance should be consistent with the guidance in Advisory Circular 120‑74A and Safety Alert for Operators 06013 and 07003. (A-07-44) Require that all 14 Code of Federal Regulations Part 91K, 121, and 135 operators install on their aircraft cockpit moving map displays or an automatic system that alerts pilots when a takeoff is attempted on a taxiway or a runway other than the one intended. (A-07-45) Require that all airports certificated under 14 Code of Federal Regulations Part 139 implement enhanced taxiway centerline markings and surface painted holding position signs at all runway entrances. (A-07-46) Prohibit the issuance of a takeoff clearance during an airplane’s taxi to its departure runway until after the airplane has crossed all intersecting runways. (A-07-47) Revise Federal Aviation Administration Order 7110.65, “Air Traffic Control,” to indicate that controllers should refrain from performing administrative tasks, such as the traffic count, when moving aircraft are in the controller’s area of responsibility. (A-07-48)

[ANALYSIS] Preflight Activities and Actions During the Taxi 2.2.2.2  Because the availability of cues and aids for the pilots’ wayfinding task was not a factor in this accident, the Safety Board examined the crew’s actions during the preflight and taxi phases of the flight’s operation to identify possible reasons for the error. The flight crew proceeded from the Comair operations center to the air carrier ramp area, where two Comair CRJ airplanes were located. The crew initially boarded the wrong airplane,162 even though the tail number of the airplane to be used for the flight 158  Such communications would have been consistent with good CRM (while at towered airports) and company policy. In addition, this behavior would have been consistent with reports of the pilots’ past performance. 159  The SkyWest first officer stated that he became “momentarily confused” about the airplane’s orientation when he looked up while crossing runway 26 but that he was able to reorient himself after identifying the holding position sign for runway 22. 160  The SkyWest and American Eagle pilots did not recall seeing the low-profile barriers on taxiway A north of runway 26. The pilots’ lack of recall in this area indicates that the barriers were not significant elements in the pilots’ wayfinding and that the taxiway A closure did not encumber their navigation to runway 22. The pilots’ lack of recall does not indicate that the barriers were inconspicuous or difficult to see when approaching or crossing runway 26. In fact, observations in a CRJ-100 airplane after the accident indicated that the barriers would have been clearly visible to the pilots during the taxi. The pilots’ failure to recall these barriers is not unusual because people can have difficulty remembering things in their environment with which they did not interact. 161  After runway 26 was reopened (November 2006), two events occurred in which air traffic controllers informed pilots that their airplanes were on that runway instead of runway 22 (see section 1.10.4). However, an interview with the pilot involved in the first event indicated that he had stopped for preflight activities, was aware of the airplane’s position, and did not intend to depart on runway 26. An interview with the captain of the flight involved in the second event indicated that he became temporarily confused about the location of runway 22 but that the airplane was never lined up to take off from runway 26. An interview with the first officer of this flight indicated that he and the captain were never confused about the correct runway for takeoff. 162  The Safety Board does not know which pilot was the first one to board the wrong airplane. Analysis National Transportation Safety Board A I R C R A F T Accident Report 64 was included in the flight release paperwork, and started its APU. Although these actions likely consumed a portion of the crew’s available time at the gate, CVR evidence and interviews with ground personnel indicated that neither pilot appeared to be rushed or hurried as he completed required tasks.

[CONCLUSIONS] The controller did not detect the flight crew’s attempt to take off on the wrong runway 13. because, instead of monitoring the airplane’s departure, he performed a lower-priority administrative task that could have waited until he transferred responsibility for the airplane to the next air traffic control facility. The controller was most likely fatigued at the time of the accident, but the extent 14. that fatigue affected his decision not to monitor the airplane’s departure could not be determined in part because his routine practices did not consistently include the monitoring of takeoffs. The Federal Aviation Administration’s operational policies and procedures at the 15. time of the accident were deficient because they did not promote optimal controller monitoring of aircraft surface operations. The first officer’s survival was directly attributable to the prompt arrival of the first 16. responders; their ability to extricate him from the cockpit wreckage; and his rapid transport to the hospital, where he received immediate treatment. The emergency response for this accident was timely and well coordinated. 17. A standard procedure requiring 14 18. Code of Federal Regulations Part 91K, 121, and 135 pilots to confirm and cross-check that their airplane is positioned at the correct runway before crossing the hold short line and initiating a takeoff would help to improve the pilots’ positional awareness during surface operations. The implementation of cockpit moving map displays or cockpit runway alerting 19. systems on air carrier aircraft would enhance flight safety by providing pilots with improved positional awareness during surface navigation. Enhanced taxiway centerline markings and surface painted holding position 20. signs provide pilots with additional awareness about the runway and taxiway environment. This accident demonstrates that 14 21. Code of Federal Regulations 91.129(i) might result in mistakes that have catastrophic consequences because the regulation allows an airplane to cross a runway during taxi without a pilot request for a specific clearance to do so. Conclusions National Transportation Safety Board A I R C R A F T Accident Report 105 If controllers were required to delay a takeoff clearance until confirming that an 22. airplane has crossed all intersecting runways to a departure runway, the increased monitoring of the flight crew’s surface navigation would reduce the likelihood of wrong runway takeoff events. If controllers were to focus on monitoring tasks instead of administrative tasks when 23. aircraft are in the controller’s area of operations, the additional monitoring would increase the probability of detecting flight crew errors. Even though the air traffic manager’s decision to staff midnight shifts at Blue Grass 24. Airport with one controller was contrary to Federal Aviation Administration verbal guidance indicating that two controllers were needed, it cannot be determined if this decision contributed to the circumstances of this accident. Because of an ongoing construction project at Blue Grass Airport, the taxiway identifiers 25. represented in the airport chart available to the flight crew were inaccurate, and the information contained in a local notice to airmen about the closure of taxiway A was not made available to the crew via automatic terminal information service broadcast or the flight release paperwork.

[ANALYSIS] Analysis National Transportation Safety Board A I R C R A F T Accident Report 59 instruction. Because two airplanes, SkyWest flight 6819 and American Eagle flight 882, were given the same taxi clearance and had already correctly taxied to and held short of runway  22 without any special instructions, there was no apparent reason for the controller to have suspected that the pilots would have had difficulty navigating to the departure runway.150 Taxi to Runway 2.2.1.2  From about 0603:16 to 0603:56, while the captain was taxiing the airplane and performing navigational checking activities, both pilots resumed the nonpertinent discussion that was started while the airplane was parked at the gate. (Figure 1 shows the location of the airplane along the taxi route while this conversation was occurring.) The nonpertinent conversation was not in compliance with the sterile cockpit rule required by company procedures and 14 CFR 121.542 (see section 1.17.1.3). The primary reason for the sterile cockpit rule is to ensure that the pilots’ attention is directed to operational concerns during critical phases of flight (including taxi) and is not redirected or degraded because of nonessential activities or discussion. FDR data showed that, about 0604:33, the airplane stopped on taxiway A at the hold short line for runway 26, which was about 560 feet from the intended destination— the hold short line for runway 22. During this time, the first officer was completing the before takeoff checklist. About 0605:15, the first officer advised the controller that the airplane was ready to depart, and the controller told the flight crew that the airplane should fly the runway heading and was cleared for takeoff. Neither the first officer nor the controller stated the runway number during the request and clearance for takeoff, but ATC procedures did not require them to do so. Because the flight crew believed that the airplane was at the hold short line for runway 22 at the time of the takeoff clearance (see section 2.2.2.3), the absence of a reference to runway 22 in the request and clearance for takeoff was not a factor in this accident. The 50-second timeframe during which the airplane was stopped at the runway 26 hold short line should have provided the flight crew with ample time to look outside the cockpit and determine the airplane’s position on the airport. At this position, the flight crew would have been able to see the runway 26 holding position sign, the “26” painted runway number, the taxiway A lights across runway 26, and the runway 22 holding position sign in the distance.151 FDR data showed that, about 0605:24, the captain began to taxi the airplane across the runway 26 hold short line.

[ANALYSIS] LEX is not one of those airports; during 2005, it had about 0.5 million passenger enplanements. 218  AC 150/5340-1J also stated, “installation at other airports is at the option of the airport operator. If an airport operator decides to exercise this option, the enhanced markings must be installed at all holding positions on the airport.” Analysis National Transportation Safety Board A I R C R A F T Accident Report 96 Taxi and Takeoff Clearances 2.4.4  As stated in section 2.2.1.1, 14 CFR 91.129(i) permits pilots, after receiving taxi clearance, to cross all intersecting runways along the taxi route (without stopping) except for the assigned departure runway. On July 6, 2000, the Safety Board issued Safety Recommendations A-00-67 and -68, which asked, in part, that the FAA (1) amend 14 CFR 91.129(i) to require that all runway crossings be authorized only by specific ATC clearance and (2) amend FAA Order 7110.65 to require that, for aircraft that need to cross multiple runways, air traffic controllers issue an explicit crossing instruction for each runway after the previous runway has been crossed. The Board classified these safety recommendations “Open—Unacceptable Response” on April 11, 2006. If these safety recommendations had been implemented before this accident, the controller would have been required to issue a specific taxi clearance for the airplane to cross runway 26 and then issue a specific taxi clearance for the airplane to continue taxiing to runway 22. These procedures would have provided the flight crew with better awareness of the airplane’s position along the taxi route and would have required the controller to visually observe the airplane’s position and monitor the taxi as the airplane progressed toward the departure runway. Thus, the flight crew’s surface navigation error might have been prevented. In addition, Mitre reports cited pilot and controller concerns about the adequacy of runway crossing requirements, and most of these pilots and controllers thought that it would be beneficial to safety to modify 14 CFR 91.129(i) so that it required a specific clearance for each runway crossing.219 The Safety Board concludes that this accident demonstrates that 14 CFR 91.129(i) might result in mistakes that have catastrophic consequences because the regulation allows an airplane to cross a runway during taxi without a pilot request for a specific clearance to do so. Therefore, the Safety Board reiterates Safety Recommendations A-00-67 and -68. In addition, no FAA guidance specifically prohibits issuing a takeoff clearance until all intersecting runways to the departure runway have been crossed.

[ANALYSIS] Because no other traffic was on the airport surface to pose a conflict to the airplane during its taxi, the controller would not have expected much useful information to be obtained by frequently scanning the runway environment, which would have decreased the likelihood that he was frequently looking out the tower cab windows at the runway environment during the 50-second window of opportunity. The Safety Board concludes that the controller did not notice that the flight crew had stopped the airplane short of the wrong runway because he did not anticipate any problems with the airplane’s taxi to the correct runway and thus was paying more attention to his radar responsibilities than his tower responsibilities. Postaccident observations from the LEX ATCT revealed that, from the controller’s work station at night, it was somewhat difficult to see whether the CRJ-100 demonstration airplane was located at the hold short line for runway 26, on taxiway A, or at the hold short line for runway 22 because of (1) the proximity of these locations in the controller’s visual field as a result of the new taxiway configuration198 and (2) the reduced visibility of ground texture, linear perspective, and other monocular depth cues that are useful for judging distances beyond 15 to 20 feet. Although the controller had been working at the tower for 17 years and had presumably become an expert at recognizing aircraft positions on the airport surface, the use of former taxiway A5 (redesignated as taxiway A) to reach the runway 22 threshold was new to him because the runway had been shifted and the redesignated taxiway had been in place for only 1 week (four of the controller’s shifts) 198  From the tower cab, the hold short lines for runways 22 and 26 appeared close to each other because they were separated by less than 5º of visual angle. Analysis National Transportation Safety Board A I R C R A F T Accident Report 81 at the time of the accident.199 These factors would have made it more difficult for the controller to determine the airplane’s exact location. In addition, the controller was not required to determine that the airplane had reached the departure runway before he cleared the airplane for takeoff; he was only expected to determine that the airplane was at a location that was consistent with its taxi clearance. When the flight crew had stopped the airplane at the runway 26 hold short line, the airplane was in a location that was consistent with its taxi clearance. The controller reported that he did not see the airplane stop in this position. Even if the controller had seen the airplane at that time and noticed that it was not moving, a brief scan of the runway environment would not have informed him of whether the airplane had been stopped only briefly or for a longer period of time. Nevertheless, the controller could have detected that the airplane had stopped short of the wrong runway if he had been monitoring the airplane’s progress along the taxi route.

[ANALYSIS] The first officer was preparing for an opportunity to upgrade to captain and was described as someone who would have made a good captain because of his adherence to standard operating procedures. 155  Even though the FAA does not consider the SEA and MIA incidents (discussed in section 1.18.4) to be runway incursions, the Safety Board does consider these incidents to be runway incursions. Analysis National Transportation Safety Board A I R C R A F T Accident Report 62 The Safety Board examined possible reasons why the flight crew stopped at the incorrect hold short line and attempted to take off from the incorrect runway, as discussed in sections 2.2.2.1 through 2.2.2.4. Available Cues and Aids for Wayfinding 2.2.2.1  Pilots integrate many sources of wayfinding information, either alone or in combination, to establish their position and navigate to their intended destination on the airport. Wayfinding information sources include geographic knowledge of standard airport surface features that are common to all airports (such as terminal buildings, ramp areas, taxiways, and runways), knowledge of standard conventions for marking and identifying airport surface features, airport charts, and taxi instructions from ATC. Because both pilots were experienced, they should have been knowledgeable about basic airport geographic features and signage and surface marking standards (the Safety Board did not note any signage or surface markings at LEX that were not in compliance with FAA regulations), and they should have been skilled at interpreting airport charts. In addition to their knowledge about the specific route (taxiway A to runway 22), multiple external cues and features were available to the pilots to support their navigation to runway 22, and no evidence indicated that their view from the windscreen was obstructed. Observations from a CRJ-100 airplane after the accident demonstrated that, during nighttime conditions, taxiway location signs were visible along the full length of the taxi route, a runway 26 holding position sign adjacent to the runway 26 hold short line was visible, and the runway 26 numbers were visible. Also, evidence indicated that, at the time of the accident, the signs identifying the critical features along the taxi route (that is, the runway 26 holding position sign, the taxiway A extension across runway 26, and the runway 22 holding position sign) were illuminated and would have been visible to both pilots.156 The flight crewmembers had resources available to them within the cockpit to support their navigation to runway 22, including the Jeppesen airport chart. Even though discrepancies existed between the airport chart and the external cues available to the pilots because of an ongoing construction project at the airport,157 the chart depicted the paved taxiway and runway surfaces at the time of the accident. Another available resource within the cockpit was the instrumentation, including the heading bug, which had been set to 227º to correspond to the magnetic heading for runway 22.

[ANALYSIS] Therefore, the Safety Board reiterates Safety Recommendations A-00-67 and -68. In addition, no FAA guidance specifically prohibits issuing a takeoff clearance until all intersecting runways to the departure runway have been crossed. On January 4, 2007, the LEX air traffic manager issued a notice that stated that controllers at the tower were not to issue takeoff clearances for runway 22 until the departing airplanes were observed to have completely crossed runway 26.220 Such guidance would benefit other airports with intersecting runways. On June 1, 2007, the FAA issued Notice N JO 7110.468 to amend the required phraseology for issuing departure instructions. According to this notice, a controller has to specifically clear an airplane across all intervening runways before issuing a takeoff clearance. However, this guidance does not instruct controllers to wait until an airplane has crossed the runways before issuing the takeoff clearance. 219  (a) Reports by Airline Pilots on Airport Surface Operations: Part 2. Identified Problems and Proposed Solutions for Surface Operational Procedures and Factors Affecting Pilot Performance, Technical Report No. MTR94W0000060.v2, McLean, Virginia: Mitre Corporation, 1994; and (b) Reports by Air Traffic Control Tower Controllers on Airport Surface Operations: The Causes and Prevention of Runway Incursions, Technical Report No. MTR98W0000033, McLean, Virginia: Mitre Corporation, 1998. 220  In addition, in its July 17, 1989, letter transmitting Safety Recommendation A-89-74 (see section 1.18.3.1), the Safety Board noted that the controllers at HOU were required to observe airplanes cross the approach end of runway 17 before issuing a clearance for takeoff for runway 12. This requirement was the result of two pilot deviation events in early 1989 that involved departures of U.S. air carrier airplanes from the wrong runway at the airport. Analysis National Transportation Safety Board A I R C R A F T Accident Report 97 The Safety Board concludes that, if controllers were required to delay a takeoff clearance until confirming that an airplane has crossed all intersecting runways to a departure runway, the increased monitoring of the flight crew’s surface navigation would reduce the likelihood of wrong runway takeoff events. Therefore, the Safety Board believes that the FAA should prohibit the issuance of a takeoff clearance during an airplane’s taxi to its departure runway until after the airplane has crossed all intersecting runways.

[ANALYSIS] In addition, the angle from the runway 26 hold short line on taxiway A to runway 26 was the same as the angle from the runway 22 hold short line on former taxiway A (north of runway 8/26) to runway 22. Also, the taxiway A centerline split into three lines after the runway 26 hold short line. These extended taxiway centerlines led onto the closed portion of taxiway A, across runway 26 to runway 22, and onto the runway 26 centerline (a lead-on/off line). The presence of a lead-on/off line from a taxiway directly to a runway could have supported the captain’s perception that the airplane had arrived at the departure runway. In addition, as stated in section 2.2.1.1, the first officer told the captain during his takeoff briefing, “lights are out all over the place,” in reference to observations he made while landing on runway 22 about 0140 on the day before the accident. The first officer’s statement might have contributed to the captain’s perception that the airplane was taxiing onto runway 22 because he might have anticipated a dark runway environment. At the 165  R. Conejo and C.D. Wickens, The Effects of Highlighting Validity and Feature Type on Air‑to‑Ground Target Acquisition Performance, Technical Report ARL-97-11, Savoy, Illinois: University of Illinois Aviation Research Lab, 1997. 166  C.D. Wickens and J.S. McCarley, Attention-Situation Awareness (A-SA) Model of Pilot Error, Technical Report ARL-01-13/NASA-01/06, Savoy, Illinois: University of Illinois Aviation Research Lab, 2001. 167  Jason S. McCarley, Margaret J. Vais, Heather Pringle, Arthur F. Kramer, David E. Irwin, and David L. Strayer, “Conversation Disrupts Change Detection in Complex Traffic Scenes,” Human Factors, vol. 46, no. 3, pages 424-436, 2004. Analysis National Transportation Safety Board A I R C R A F T Accident Report 66 least, this statement reduced the significance of the first officer’s subsequent statement, “[that] is weird with no lights” (to which the captain responded “yeah”), as the airplane rolled down runway 26. There are well-known psychological concepts associated with perception and decision-making that can allow a person’s mistaken assessment to persist. For example, confirmation bias occurs when people seek out or observe elements in their environment that support their perception. Specifically, confirmation bias results from a tendency for people to primarily seek out confirming evidence of a belief while spending less effort to seek out negative evidence that can disconfirm the belief.168 Confirmation bias can cause a person to persist in holding an incorrect belief despite the availability of contradictory evidence.

[ANALYSIS] A NOTAM, which was included in the flight release paperwork, and the ATIS information broadcast, which the flight crew received, indicated Analysis National Transportation Safety Board A I R C R A F T Accident Report 76 that runway 27 was closed. Also, the flight crew reported that the airport charts were out and available. As the airplane proceeded down runway 27, the airplane’s nose wheel light illuminated a truck flashing its lights. The captain and the first officer stated that they observed the truck at the same time. Simultaneously, the controller queried the flight crew to determine whether the airplane was on runway 30. The flight crew rejected the takeoff, and the airplane continued uneventfully to its destination. The Comair flight 5191 accident and other wrong runway takeoff events demonstrate that all pilots are vulnerable to this and other types of surface navigation errors. Even when navigation tasks are straightforward and simple, there is a potential for a catastrophic outcome resulting from human error if available cues are not observed and considered during taxi and the airplane’s position is not cross-checked at the intended runway. Systemwide interventions, such as improved airport standard markings and technologies to improve pilots’ positional awareness while navigating airport surfaces, can help protect against human error during airport surface operations by providing a redundant display of critical information. Such interventions are further discussed in section 2.4. In addition, the Safety Board is concerned about the breakdown in sterile cockpit discipline in the sequence of events leading to this accident. The pilots’ actions with regard to sterile cockpit procedures, specifically, their nonpertinent conversation while the captain was taxiing the airplane, were uncharacteristic with past reports. A first officer who flew with the captain 2 days before the accident described the captain’s sterile cockpit discipline as very good and stated that, when a newly trained first officer was occupying the cockpit jumpseat during one of the flight legs, the captain briefed him about sterile cockpit discipline. The first officer was also described as having good sterile cockpit discipline. On February 7, 2006, the Safety Board issued Safety Recommendation A-06-7, which was the result of the Board’s investigation of the Corporate Airlines flight 5966 accident.

[ANALYSIS] At this position, the flight crew would have been able to see the runway 26 holding position sign, the “26” painted runway number, the taxiway A lights across runway 26, and the runway 22 holding position sign in the distance.151 FDR data showed that, about 0605:24, the captain began to taxi the airplane across the runway 26 hold short line. FDR data also showed that, about 0605:41, the airplane began to turn onto runway 26, and the CVR showed that, about 0605:46, the first officer completed the lineup checklist. 150  In accordance with FAA Order 7110.65, “Air Traffic Control,” paragraph 3-7-2, the controller would have been required to provide turn-by-turn directions to the departure runway if the flight crew had so requested. 151  During the 50-second timeframe, the controller did not query the flight crew regarding why the airplane was stopped at the hold short line for runway 26. The controller’s actions during the taxi and attempted takeoff sequence are discussed in section 2.2.3. Analysis National Transportation Safety Board A I R C R A F T Accident Report 60 Takeoff Roll 2.2.1.3  About 0605:58, the captain transferred control of the airplane to the first officer, stating, “all yours,” to which the first officer acknowledged, “my brakes, my controls.” At this time, the captain would have switched his attention from outside to inside the cockpit, and the first officer would have switched his attention from inside to outside the cockpit. About 2 seconds later, the airplane was aligned with the centerline for runway 26. The CVR recording showed that the flight crew had referred to runway 22 as the departure runway multiple times before takeoff, and FDR data showed that the pilots’ heading bugs were set to 227º, which was consistent with the magnetic heading for runway 22. The Safety Board concludes that the captain and the first officer believed that the airplane was on runway 22 when they taxied onto runway 26 and initiated the takeoff roll. About 0606:16, the first officer stated, “[that] is weird with no lights,” to which the captain responded, “yeah.” At that time, the airplane was passing through the intersection of runway 26 with runway 22. About 0606:24, the captain called the 100‑knot airspeed check. About the same time, the airplane accelerated beyond the maximum airspeed that would have allowed the airplane to remain on the available runway if the flight crew had rejected the takeoff and used maximum braking.152 At 0606:31.2, the captain called, “V one, rotate,” followed immediately by his exclamation, “whoa.” The aircraft performance study for this accident showed that, at the time of the VR callout, the airplane was 236 feet from the end of the runway.

[ANALYSIS] Two two’s a short taxi.” Although the CVR did not record either pilot explicitly referencing the airport chart, this statement is consistent with the first officer examining the chart because no specific taxi instructions had been provided to the flight crew. Also, the number of times that each crewmember had previously arrived at or departed from LEX was likely not sufficient to allow either one to have memorized taxiway identifiers and routes. During the takeoff briefing, the first officer stated that the runway end identifier lights were out and then commented, “came in the other night it was like … lights are out all over the place.” The first officer was referring to observations he made on a repositioning flight that landed on runway 22 about 0140 on the day before the accident. (The right runway edge lights after the intersection of runway 22 with runway 26 were out at that time.) The first officer did not brief that the taxi to runway 22 required crossing runway 26. The Safety Board was unable to determine why this information was not included in the first officer’s briefing. It is possible that the simplicity of the taxi and the use of only one taxiway might have led him to assume that it was unnecessary to include this additional information in his briefing. During postaccident interviews, other pilots indicated that they would brief this “short taxi”148 in a similar manner. No evidence indicated that the pilots were unaware of the need to cross runway 26 to arrive at runway 22. About 0602:01, the first officer notified the controller that the airplane was ready to taxi. The controller then instructed the flight crew to taxi the airplane to runway 22. Title 14 CFR 91.129(i), “Takeoff, Landing, Taxi Clearance,” permits pilots instructed to taxi to a point on the airport surface to cross all intersecting runways along the taxi route (without stopping) except for the assigned departure runway.149 Thus, the controller’s clearance for the airplane to taxi to runway 22 complied with 14 CFR 91.129(i), and the first officer’s response of “taxi two two” was an appropriate acknowledgment of the taxi 146  Comair’s high threat taxi procedures are used at those airports with an operating environment that presented “exceptional” hazards to safe taxi. However, these procedures were not in place for operations at LEX at the time of the accident. 147  As stated in section 1.1, the takeoff briefing is part of the before starting engines checklist. 148  Taxiway A is used for the taxi to both runways 4 and 22, but the taxi to runway 22 is significantly shorter than the taxi to runway 4, as shown on airport charts (see appendix C). 149  This regulation is discussed further in section 2.4.4. Analysis National Transportation Safety Board A I R C R A F T Accident Report 59 instruction.

[ANALYSIS] Dismukes, “Concurrent Task Management and Prospective Memory: Pilot Error as a Model for the Vulnerability of Experts.” Proceedings of the Human Factors and Ergonomics Society 50th Annual Meeting, 2006. 164  R.K. Dismukes, G.E. Young, and R.L. Sumwalt, Cockpit Interruptions and Distractions: Effective Management Requires a Careful Balancing Act. ASRS Directline, vol. 10, pages 4-9, 1998. Analysis National Transportation Safety Board A I R C R A F T Accident Report 65 in either a slight deviation from an intended path, or becoming totally lost.”165 In addition, active attention is thought to be necessary for maintaining situational awareness, and the allocation of attention to irrelevant stimuli can degrade awareness.166 Finally, research on distractions while driving found that drivers are less likely to detect changes that have occurred in the environment when they are involved in casual conversation because being engaged in a conversation may degrade the encoding of visual information.167 The first officer initiated the nonpertinent conversation as the captain was navigating along the taxi route. The captain had the responsibility to assert both his leadership role and command authority to stop the discussion. Rather, the captain allowed the conversation to continue and participated in it. Also, instead of initiating the nonpertinent conversation, the first officer should have been monitoring the captain’s actions and independently assessing the airplane’s location along the taxi route. The Safety Board concludes that the flight crewmembers’ nonpertinent conversation during the taxi, which was not in compliance with Federal regulations and company policy, likely contributed to their loss of positional awareness. It is important to note that the CVR did not record any statement by either flight crewmember about this loss of positional awareness. Cues to Indicate a Takeoff From Runway 22 2.2.2.3  The presence of runway markings—a white centerline and side stripes—ahead of the airplane would have facilitated the captain’s perception that the airplane had arrived at the hold short line for runway 22, even though the airplane was actually at the hold short line for runway 26. In addition, the angle from the runway 26 hold short line on taxiway A to runway 26 was the same as the angle from the runway 22 hold short line on former taxiway A (north of runway 8/26) to runway 22. Also, the taxiway A centerline split into three lines after the runway 26 hold short line.

[ANALYSIS] The Safety Board was not able to determine what effect, if any, this information would have had on the circumstances of this accident. However, this possible cue would not have had the same salience as the primary cues— the airport markings and signage—that were accurate and available to the flight crew. In addition, the Board’s analysis of the accident determined that the taxi to runway 22 was relatively simple and could have been successfully conducted by using the cues and aids available to the flight crew. Thus, the Safety Board concludes that, because the information in the local NOTAM about the altered taxiway A configuration was not needed for the pilots’ wayfinding task, the absence of the local NOTAM from the flight release paperwork was not a factor in this accident. As stated in section 1.18.6, the FAA has planned initiatives to modernize the current NOTAM system. These initiatives include aligning the U.S. NOTAM system with that of ICAO by October 2007 so that U.S. NOTAM information can be processed and provided to flight crews in a more timely, accurate, complete, and traceable manner and having digital NOTAM data displayed in the cockpit in textual and graphical formats. For this accident, even though the information in the local NOTAM was not necessary for the flight crew’s successful navigation to the correct runway, such information might be necessary under different circumstances. Thus, the Safety Board is pleased that the FAA is taking a proactive role to improve the current NOTAM system. Presence of Extended Taxiway Centerline 2.6.4  One of the three centerlines that diverged from the taxiway A hold short position for runway 8/26 led to taxiway A north of runway 8/26, which had been closed and was blocked by low-profile barricades with flashing red lights. This extended taxiway centerline had not been removed at the time of the accident. Analysis National Transportation Safety Board A I R C R A F T Accident Report 102 The LEX construction plan called for the removal of taxiway A north of runway 8/26 and the associated extended taxiway centerline within 30 to 45 days of that taxiway’s closure. Because proper barricades had been put into place until such time, and because the CVR did not record any confusion between the flight crewmembers about the taxiway A configuration, the Safety Board concludes that the presence of the extended taxiway centerline to taxiway A north of runway 8/26 was not a factor in this accident. National Transportation Safety Board A I R C R A F T Accident Report 103 Conclusions 3. Findings 3.1  The captain and the first officer were properly certificated and qualified under Federal 1. regulations. There was no evidence of any medical or behavioral conditions that might have adversely affected their performance during the accident flight. Before reporting for the accident flight, the flight crewmembers had rest periods that were longer than those required by Federal regulations and company policy.

[ANALYSIS] The referenced data were provided to the Safety Board in January 2007. Analysis National Transportation Safety Board A I R C R A F T Accident Report 74 had only to follow those signs and the taxiway centerline to the runway holding position sign for runway 22. Although this task is more difficult during night conditions, when surrounding features in the environment and horizon as well as other cues are not easily detectable, the critical features along the navigation route were internally lit or were illuminated by the airplane’s external lighting system. Both pilots were described to be in good health in the days before the accident.185 Postaccident toxicological testing for both pilots did not detect the presence of alcohol186 or any substances known to affect performance. Both pilots were required to wear corrective lenses during flight operations. Evidence indicated that the captain was wearing contact lenses during the accident flight, but the Safety Board could not determine whether the first officer was wearing corrective lenses during the flight. Even if the first officer had not been wearing corrective lenses, his ability to verify the airplane’s position during the taxi to runway 22 would most likely not have been affected.187 Evidence indicated that both pilots were able to read and interpret features inside the cockpit at intermediate distances and detect objects outside the cockpit at greater distances, and the pilots were described to have normal night vision. Observations in a CRJ-100 airplane after the accident demonstrated that no objects along the taxi route would have produced temporary flash blindness188 that could have impeded the pilots’ ability to detect signage and surface markings.189 As previously stated, numerous cues were available to the flight crew to indicate that the airplane was not at the position on the airport surface that it was supposed to be. For example, when the airplane stopped at the runway 26 hold short line instead of the runway 22 hold short line, the flight crew would have been able to see the runway 26 holding position sign and painted runway numbers, the continuation of taxiway A across runway 26, and the runway 22 holding position sign. Also, when the airplane was in 185  The first officer tested positive for a low level of pseudoephedrine. This drug, at low levels, is not considered to cause impairment. Performance-based side effects of pseudoephedrine are similar to those of caffeine. CAMI does not quantify this substance unless it is above therapeutic levels. The Safety Board was not able to determine why and when the medication was taken. 186  On August 26, 2006, at 1826, the first officer purchased food and two bottles of beer at the hotel restaurant. (Section 1.17.1.6 describes the FAA’s and Comair’s alcohol use policy.) According to the CAMI toxicology report, the first officer tested negative for ethanol.

[ANALYSIS] The appearance of the runway end environment would have provided a salient cue to the flight crew that the airplane was in an extremely hazardous situation and could not remain on the ground. The airplane’s airspeed at the time of the captain’s VR callout was 131 knots, which was 11 knots below the planned VR airspeed of 142 knots (which the flight crew had briefed and entered into the airplane’s EFIS).153 Thus, the captain’s early VR callout and subsequent “whoa” exclamation indicated that he recognized that something was wrong with the takeoff. FDR data showed that, in response, the first officer pulled the control column full aft154 and that the airplane rotated at a rate of about 10º per second, which is three times the normal rotation rate. This abnormal column input showed that the first officer also recognized that something was wrong with the takeoff. Although numerous cues, including the lack of runway lighting, were available to indicate that the airplane was not on the assigned runway (see sections 2.2.2.1 and 2.2.2.3), 152  According to calculations by Bombardier, for the airplane to have stopped before the end of runway 26, maximum braking would have had to start when the airplane was at an airspeed of about 103 knots. 153  FDR data for the accident airplane’s 12 previous takeoffs indicated that rotation occurred at or after reaching the VR airspeed. 154  FDR data showed that the left and right control column inputs during the accident rotation reached 10.6º and 10.9º, respectively, and that the nominal control column input for rotation during the accident airplane’s 12 previous takeoffs was between 4º and 5º. Analysis National Transportation Safety Board A I R C R A F T Accident Report 61 the flight crew had not correctly interpreted these cues or noticed them until after it was too late to successfully abort the takeoff. The Safety Board concludes that the flight crew recognized that something was wrong with the takeoff beyond the point from which the airplane could be stopped on the remaining available runway.

[ANALYSIS] Automatic Terminal Information Service Broadcasts 2.6.2  The flight crew indicated that it had received ATIS information “alpha,” and the controller advised the flight crew when ATIS information “bravo” was current. Neither ATIS broadcast indicated that taxiway A north of runway 8/26 had been closed as part of the airport construction project. FAA Order 7110.65, paragraph 2-9-3g, “Content,” states that taxiway closures that affect the entrance or exit of active runways should be included in ATIS broadcasts. After the accident, the Safety Board determined that the taxiway closure information had been included in a sampling of ATIS broadcasts recorded between 0530 and 0630 each day between August 20 and 26, 2006. FAA Order 7110.65, paragraph 2-9-2d, “Operating Procedures,” allows controllers to supplement ATIS information during their direct communications with pilots. The CVR showed that the controller did not advise the flight crew about the status of taxiway A north of runway 8/26. Because the construction project affected the entrance and exit paths of runway 4/22, either the ATIS broadcasts or the controller’s direct communications with the pilots should have included information about the altered taxiway A configuration. The Safety Board was not able to determine what effect, if any, information about the closure of taxiway A north of runway 8/26 (either through the ATIS broadcasts or the controller’s direct communications with the flight crew) would have had on the circumstances of this accident. However, this possible cue would not have had the same salience as the primary cues—the airport markings and signage—that were accurate Analysis National Transportation Safety Board A I R C R A F T Accident Report 101 and available to the flight crew. It is also important to note that the airplane turned onto runway 26 before it would have encountered the area on taxiway A that had been closed. Thus, the Safety Board concludes that the controller’s failure to ensure that the flight crew was aware of the altered taxiway A configuration was likely not a factor in the crew’s inability to navigate to the correct runway. Nevertheless, the Board recognizes that, under different circumstances, information on an altered taxiway configuration might be necessary for a flight crew’s successful navigation on an airport and that, for those circumstances, the crew would need to receive complete, accurate, and timely ATIS information. Local Notice to Airmen 2.6.3  The local NOTAM that indicated that taxiway A north of runway 8/26 was closed was not included in the Comair flight planning system or the flight release paperwork for the accident flight because the company determined that the information in the local NOTAM did not affect safety of flight. The Safety Board was not able to determine what effect, if any, this information would have had on the circumstances of this accident. However, this possible cue would not have had the same salience as the primary cues— the airport markings and signage—that were accurate and available to the flight crew.

[ANALYSIS] These interventions can also help prevent runway incursions, which is an issue on the Safety Board’s list of Most Wanted Transportation Safety Improvements. 210  This definition was cited in the Safety Board’s 1981 special study on cabin safety in large transport aircraft. According to the study, the definition (1) was developed using aviation crash injury research by Cornell University and aviation safety engineering and research by the Flight Safety Foundation and (2) was used in the Aircraft Crash Survival Design Guide, which was prepared by the U.S. Army Research and Technology Laboratories along with other Federal agencies. The definition has been used by the Board since that time to determine the survivability of accidents. In addition, the definition appeared in the Board’s 2001 safety report on the survivability of accidents involving Part 121 U.S. air carrier operations from 1983 to 2000. Analysis National Transportation Safety Board A I R C R A F T Accident Report 91 Flight Deck Procedures 2.4.1  Well-designed flight deck procedures can be an effective countermeasure against surface operation errors. After this accident, the Safety Board recognized the need to improve industry standards for confirming an airplane’s position at the departure runway before takeoff and, on December 12, 2006, issued Safety Recommendation A-06‑83. This recommendation asked that the FAA require all Part 121 operators to establish procedures requiring all crewmembers on the flight deck to positively confirm and cross-check the airplane’s location at the assigned departure runway before crossing the hold short line for takeoff. On April 16, 2007, the FAA issued SAFO 07003, “Confirming the Takeoff Runway,” in response to Safety Recommendation A-06-83. According to the SAFO, its purpose is to emphasize the importance of implementing standard operating procedures and training for flight crews to ensure that an airplane is at the intended runway. SAFO 07003 was aimed at directors of safety, directors of operations, fractional ownership program managers, trainers, and pilots. The SAFO stated that pilots should positively confirm and cross-check the takeoff runway and the airplane’s location at the assigned departure runway before crossing the hold short line and while in the takeoff position. The SAFO further stated that airplane-specific standard operating procedures should be established, implemented, and supported by pilot training that uses all available resources to confirm and cross-check an airplane’s position. The SAFO mentioned that these resources included the HSIs, which can confirm that an airplane’s position is where the flight crew intended, and air traffic controllers, who can help confirm an airplane’s position during taxi or at a hold short line. The Safety Board is encouraged that the FAA is providing renewed emphasis about the importance of cross-checking and confirming an airplane’s position on a runway.

[ANALYSIS] Even if the controller had seen the airplane at that time and noticed that it was not moving, a brief scan of the runway environment would not have informed him of whether the airplane had been stopped only briefly or for a longer period of time. Nevertheless, the controller could have detected that the airplane had stopped short of the wrong runway if he had been monitoring the airplane’s progress along the taxi route. Critical Window During Which an Administrative Task Was Performed 2.2.3.2  After the LEX controller cleared Comair flight 5191 for takeoff, he told American Eagle flight 882 to contact the Indianapolis ARTCC. According to the ATC transcript, the handoff of the American Eagle flight to the Indianapolis ARTCC occurred about 0605:40. About that time, Comair flight 5191, the only airplane for which the controller had responsibility, was turning onto runway 26 and had not yet deviated from the issued taxi clearance. The controller’s next active task would be to establish contact with the Comair flight and provide departure services (radar tasks), but he would likely not have expected to perform this task for about 1 minute.200 This 1-minute pause in active control tasks afforded the controller greater flexibility in terms of his allocation of attention. The controller stated that, after the handoff of the American Eagle flight to the Indianapolis ARTCC, he began the combined traffic count, which was an administrative record-keeping task.201 The standard operating procedure at LEX was to perform the traffic count on an hourly basis. However, the controller stated that he normally accumulated flight progress strips throughout the night and performed the traffic count once toward the end of his shift. The controller estimated that the traffic count would take 2 to 5  minutes to complete.202 The controller was expecting to be relieved by the incoming day shift controller at 0630, so he most likely wanted to complete the traffic count by that time. However, 199  When taxiway A north of runway 8/26 was used to reach the runway 22 threshold, the hold short lines for runways 22 and 26 were farther apart in the controller’s visual field. 200  The time between flight crew acknowledgment of the takeoff clearance and the controller’s acknowledgment of radar contact was 41 and 78 seconds for the SkyWest and American Eagle departures, respectively. 201  When the tower was staffed with more than one controller, the controller at the radar data position recorded the radar traffic count, and the controller at the clearance delivery position recorded the tower traffic count. When the midnight shift was staffed with one controller, that controller performed a combined radar and tower traffic count. 202  As a result, the controller must have expected that he would have to interrupt this administrative task to provide radar services to the Comair flight.

[ANALYSIS] For example, confirmation bias occurs when people seek out or observe elements in their environment that support their perception. Specifically, confirmation bias results from a tendency for people to primarily seek out confirming evidence of a belief while spending less effort to seek out negative evidence that can disconfirm the belief.168 Confirmation bias can cause a person to persist in holding an incorrect belief despite the availability of contradictory evidence. For the flight crew, confirmation bias was in place not only at the hold short line for runway 26 but also during the initial acceleration down the runway because the crew did not evaluate evidence that would contradict the airplane’s position on the airport surface at the time. There were cues available to the flight crew that were not consistent with a taxi onto runway 22. These cues included the runway holding position sign for runway 26, the 75-foot painted width of runway 26 (versus the 150-foot width of runway 22),169 and the absence of runway edge lights and precision runway markings (such as threshold markings and touchdown zone markings) on runway 26. However, once the flight crewmembers had navigated to what they believed was the correct runway, they were likely no longer giving strong weight to contradictory information as a result of confirmation bias. On July 17, 1989, the Safety Board issued Safety Recommendation A-89-74, which asked the FAA to ensure that the operations manuals of all Part 121 and 135 air carriers require flight crews to cross-check the heading indicator with the runway heading when the airplane is aligned with the runway for takeoff. The Board classified Safety Recommendation A-89-74 “Closed—Acceptable Action” on December 11, 1990, after the FAA revised ACOB 8-85-1 to address the need for pilots to cross-check the heading indicator with the runway heading before takeoff. These bulletins, however, are not mandatory, and, during this investigation, the Board determined that Comair and other Part 121 operators did not have procedures for positively verifying that an airplane was aligned on the correct departure runway.170 FDR data showed that, at some point before the start of the recording, the pilots’ heading bugs had been set to 227º to correspond to the magnetic heading for runway 22. 168  The Safety Board notes that confirmation bias has multiple definitions. This report refers to confirmation bias as a phenomenon that occurs automatically (that is, without conscious intervention or intent) at the perceptual level. 169  Postaccident observations from a CRJ-100 airplane indicated that the reduced available width of runway 26 was clearly visible at night because the side stripes were brightly illuminated by the airplane’s external lighting system. 170  As a result of this finding and the finding that that some Part 121 operators (including Comair) did not provide guidance to their pilots about conducting takeoffs at night on unlighted runways, the Safety Board issued Safety Recommendations A-06-83 and -84 on December 12, 2006 (see sections 1.18.3.1 and 2.4.1).

[ANALYSIS] For example, in January 2014, a Boeing 737 landed at the wrong airport in Branson, Missouri, in night VMC. The flight crew expected that the visually identified airport and runway were the intended destination and did not reference cockpit displays to verify the airport and runway. As a result, the airplane landed on runway 12 at M. Graham Clark Downtown Airport instead of runway 14 at Branson Airport. Also, in November 2013, a Boeing 747 landed at the wrong airport in Wichita, Kansas, in night VMC due to the flight crew’s expectation that the observed runway lights were from the intended landing runway at McConnell Air Force Base. Instead, the airplane landed at Colonel James Jabara Airport on a runway that was one-half the length of the intended landing runway. For both of these cases, cues that indicated the flight crew’s mistaken perception were available; however, those cues were not effectively used because the crewmembers’ expectation bias outweighed the available conflicting cues.101 For this incident, lighting aids generally associated with runways were not present on taxiway C. Specifically, although the flight crew perceived the taxiway to be the intended runway, the taxiway did not have a precision approach path indicator, touchdown zone lights, full-length edge lights, and approach lights.102 However, the absence of these normally conspicuous features of a runway would have been difficult for the flight crewmembers to recognize because of their expectation bias and the inherent difficulty detecting omissions in the environment (the latter of which could have been mitigated if the flight crew had briefed the runway 28L closure). In addition, features present along taxiway C were inconsistent with it being a runway. For example, although the presence of centerline lights along the full surface length was a cue that was consistent 101 For more information, see NTSB incident numbers DCA14IA037 and DCA14IA016, respectively. 102 Air Canada’s stabilized approach criteria for a visual approach included vertical tracking on approximately a 3° glidepath and using a visual approach slope indicator. Postincident interviews and airplane track data suggested that the captain used the precision approach path indicator located to the left of runway 28R for glidepath information, which was intended for airplanes approaching that runway. The availability of this glidepath information while the airplane was aligned with taxiway C would have supported the flight crew’s expectation that the airplane was aligned with runway 28R. NTSB Aircraft Incident Report 51 with a runway, the taxiway centerline lights were green, as shown in figure 4.103 (Runway centerline lights are white.) Also, flashing yellow in-pavement guard lights were present on taxiway C (also shown in figure 4), which would not have been present on a runway surface because the guard lights were designed to prevent a taxiing airplane from crossing onto a runway. During postincident interviews, the flight crewmembers recalled seeing specific color cues, including the green taxiway centerline lights.104 However, the flight crew continued the approach despite this conflicting cue.

[CONCLUSIONS > FINDINGS] Although the use of line up and wait (LUAW) procedures during single-person air traffic control operations was not a factor in this incident, the tower controllers should have delayed consolidating local and non-local control positions until LUAW procedures were no longer needed. NTSB Aircraft Incident Report 68 18. If an airplane were to align with a taxiway, an automated airport surface detection equipment alert would assist controllers in identifying and preventing a potential taxiway landing as well as a potential collision with aircraft, vehicles, or objects that are positioned along taxiways. 19. Increased conspicuity of runway closure markers, especially those used in parallel runway configurations, could help prevent runway misidentification by flight crews while on approach to an airport. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this incident was the flight crew’s misidentification of taxiway C as the intended landing runway, which resulted from the crewmembers’ lack of awareness of the parallel runway closure due to their ineffective review of notice to airmen (NOTAM) information before the flight and during the approach briefing. Contributing to the incident were (1) the flight crew’s failure to tune the instrument landing system frequency for backup lateral guidance, expectation bias, fatigue due to circadian disruption and length of continued wakefulness, and breakdowns in crew resource management and (2) Air Canada’s ineffective presentation of approach procedure and NOTAM information. NTSB Aircraft Incident Report 69

[ANALYSIS] The flight crewmembers’ training records indicated no issues with identifying airport surfaces, flying stabilized approaches, and flying visual approaches. The incident occurred in night VMC, and no evidence indicated any obstructions or glare in the cockpit that would have affected the flight crew’s view outside of the cockpit windows. However, the flight crewmembers were unable to identify the runway 28R surface (despite the presence of approach and runway lighting) and instead aligned the airplane with parallel taxiway C. Also, neither crewmember recognized that the airplane was not aligned with the intended landing runway until the airplane was over the airport surface, at which time the flight crew initiated a low-altitude go-around. Sections 2.3.1 through 2.3.3 discuss reasons for the flight crew’s alignment error and the factors that led to the eventual recognition of this error, and section 2.3.4 discusses the mitigation of such errors. 2.3.1 Flight Crew Awareness of Runway Closure The flight crew had opportunities before the approach to learn about the runway 28L closure. The first opportunity occurred before the flight when the crewmembers received the flight release. Both crewmembers stated that they reviewed NOTAMs in the flight release. However, the first officer stated that he could not recall reviewing the specific NOTAM that addressed the runway 28L closure. Also, even though the captain stated that he saw the runway closure information, his actions in misaligning the airplane demonstrated that he did not recall that information when it was needed, and he thought that runway 28R was runway 28L. The second opportunity occurred in flight during the crewmembers’ preparations for the approach to runway 28R. Both crewmembers recalled reviewing ATIS information Quebec, which they received via an ACARS message in the cockpit, but neither crewmember recalled reviewing the specific NOTAM in the ATIS information that described the runway 28L closure. Because the flight crewmembers either did not review or could not recall the information about the runway 28L closure, they expected to see two parallel runways while on approach to SFO and further expected that they would need to fly the approach to the right-side surface.95 The flight crew’s recent experience flying into SFO would have reinforced these expectations. For example, when the first officer flew into SFO 2 nights before the incident, the airplane used for that flight landed on runway 28R at 2305, which was about 18 minutes before runway 28L was closed. Also, the captain stated that he had never seen runway 28L “dark” and that none of his previous landings at SFO occurred when a runway was closed. A runway closure marker with a flashing white “X” was placed at the threshold of runway 28L to indicate the runway closure.

[ANALYSIS] During postincident interviews, the flight crewmembers recalled seeing specific color cues, including the green taxiway centerline lights.104 However, the flight crew continued the approach despite this conflicting cue. Given that the general outline of airplane lights along taxiway C (in a straight line) had likely confirmed the crew’s expectation that the right-side surface was a runway, the omission of conflicting color cues in the crew’s assessment of the runway environment was consistent with the effects of expectation bias.105 The captain of DAL521 (the flight that immediately preceded the ACA759 into SFO) provided a similar assessment during postincident interviews. Specifically, the DAL521 captain stated that the airplane lights on taxiway C gave the impression that that surface could have been a runway.106 Although multiple cues were available to the flight crew to distinguish runway 28R from taxiway C, sufficient cues also existed to confirm the crew’s expectation that the airplane was aligned with the intended landing runway. As a result, once the airplane was aligned with what the flight crewmembers thought was the correct landing surface, they were likely not strongly considering contradictory information. The NTSB concludes that the cues available to the flight crewmembers to indicate that the airplane was aligned with a taxiway were not sufficient to overcome their belief, as a result of expectation bias, that the taxiway was the intended landing runway. 2.3.2.2 Flight Crew Recognition of Misalignment The captain stated that, while on final approach, he noticed lights going across what he thought was the runway 28R surface; this description is consistent with the in-pavement guard lights on taxiway C. Despite this cue indicating that the airplane was aligned with a taxiway, the captain’s expectation bias continued because of his assumption that the lights were associated with an airplane on the runway surface. The captain then asked the first officer to verify with the controller that the runway was clear. When the first officer looked up after prolonged heads-down time during the approach (see section 2.2.3), the airplane was lined up with the taxiway. However, the first officer presumed that the airplane was aligned with runway 28R due, in part, to his expectation that the captain would align the airplane with the runway, and the first officer did not 103 The NTSB is currently investigating a December 2017 incident involving a Horizon Air Bombardier Q400 that was attempting to land on runway 6 at Pullman-Moscow Regional Airport, Pullman, Washington, at night. The runway was dark because its lights were out of service, and the airplane landed on the taxiway that was parallel to the runway despite the illumination of blue taxiway edge lights, which the flight crew perceived as dim white runway edge lights.

[CONCLUSIONS > FINDINGS] NTSB Aircraft Incident Report 67 9. Although the notice to airmen about the runway 28L closure appeared in the flight release and the aircraft communication addressing and reporting system message that were provided to the flight crew, the presentation of the information did not effectively convey the importance of the runway closure information and promote flight crew review and retention. 10. The cues available to the flight crewmembers to indicate that the airplane was aligned with a taxiway were not sufficient to overcome their belief, as a result of expectation bias, that the taxiway was the intended landing runway. 11. Multiple salient cues of the surface misalignment were present as the airplane approached the airport seawall, and one or more of these cues likely triggered the captain’s initiation of a go-around, which reportedly occurred simultaneously with the first officer’s call for a go-around. 12. The captain and the first officer were fatigued during the incident flight due to the number of hours that they had been continuously awake and circadian disruption, which likely contributed to the crewmembers’ misidentification of the intended landing surface, their ongoing expectation bias, and their delayed decision to go around. 13. Current Canadian regulations do not, in some circumstances, allow for sufficient rest for reserve pilots, which can result in these pilots flying in a fatigued state during their window of circadian low. 14. Flight safety would be enhanced if airplanes landing at primary airports within Class B and Class C airspace were equipped with a cockpit system that provided flight crews with positional awareness information that is independent of, and dissimilar from, the current instrument landing system backup capability for navigating to a runway. 15. Although the investigation into this incident identified significant safety issues, cockpit voice recorder information, had it been available, could have provided direct evidence regarding the flight crew’s approach preparation, cockpit coordination, perception of the airport environment, and decision-making. 16. Once the flight crewmembers perceived lights on the runway, they decided to contact the controller to ask about the lights; however, their query was delayed because of congestion on the tower frequency, which reduced the time available for the crewmembers to reconcile their confusion about the lights with the controller’s confirmation that the runway was clear. 17. Although the use of line up and wait (LUAW) procedures during single-person air traffic control operations was not a factor in this incident, the tower controllers should have delayed consolidating local and non-local control positions until LUAW procedures were no longer needed. NTSB Aircraft Incident Report 68 18.

[ANALYSIS] Given the current specifications for the runway closure marker, these flight crews likely did not see the “X” because their airplanes were not aligned with runway 28L. This situation highlights the need for a surface-based system with conspicuous and unambiguous visual cues that clearly indicate when a runway is closed, even if an approaching airplane is not aligned with the closed runway. Such a system would provide redundancy in case a flight crew does not review or retain runway closure information presented in NOTAMs or ATIS broadcasts/messages. Such a system would also be especially critical in situations involving the closure of a runway at an airport with parallel landing and taxiway surfaces (such as SFO). As previously discussed, cockpit mitigations (such as the use of navigational aids to back up a visual approach to the intended landing runway, as used by the DAL521 flight crew, and a system to provide additional positional awareness information) and ATC systems to alert controllers about a potential runway misalignment or taxiway alignment can provide multiple layers of protection to prevent a flight crew surface alignment error. In addition, when closed runways are marked conspicuously, the view outside the cockpit (during VMC) of an airport environment can assure flight crews that an airplane is correctly aligned with a parallel landing runway. If the incident flight crewmembers had observed the runway 28L closure marker early in the approach, their mistaken perception of the airport environment and alignment with taxiway C might not have occurred. Further, if the runway 28L closure marker had captured the incident flight crew’s attention later in the approach, that information might have been sufficient for the crewmembers to detect their mistaken perception and respond to the situation before the airplane reached the seawall. The NTSB is not aware of any operational or human factors research to improve the conspicuity of L-893 runway closure markers during the 31 years since the FAA issued a technical report about its research on runway closure markers (other than subsequent research on the use of light-emitting diode bulbs). During that time, widespread advances in lighting and control 126 The NTSB investigated a September 25, 2001, incident in which a Boeing 757 took off from a closed runway at Denver International Airport, Denver, Colorado. A system failure affected the availability of a NOTAM about the runway closure, and a controller cleared the airplane to take off from the closed runway. After that incident, the flight crewmembers stated that they were unaware that the runway was closed, and the captain did not recall if the runway closure information was included in the ATIS broadcast. As a result of the incident, the NTSB recommended that the FAA “require the use of physical devices or other means to clearly indicate to flight crews of arriving and departing aircraft that a runway is closed” (A-03-5). In response, the FAA issued AC 150/5370-2F, “Operational Safety on Airports During Construction,” which indicated that airports should use physical devices or other means to indicate to flight crews that a runway is closed.

[ANALYSIS] The runway was dark because its lights were out of service, and the airplane landed on the taxiway that was parallel to the runway despite the illumination of blue taxiway edge lights, which the flight crew perceived as dim white runway edge lights. For more information, see NTSB incident number DCA18IA081. 104 Taxiway lighting at YYZ, where the flight crew was based, was similar to the lighting on SFO taxiway C; both airports had green taxiway centerline lights and blue taxiway edge lights. 105 The airplanes’ wingtip navigation lights were red (left wing) and green (right wing), and the runway edge lighting was white. 106 The DAL521 captain also stated that, because the lights from the airplanes located on taxiway C were in a straight line, the airplane lights could have been perceived as centerline lights, which could confuse a flight crew. NTSB Aircraft Incident Report 52 immediately recognize that the surface ahead was not the intended landing runway. When the first officer contacted the controller, the airplane was about 4,000 ft (0.66 nm) from the airport seawall. As ACA759 continued to approach the airport seawall, the flight crew of the second airplane on taxiway C, PAL115, saw that ACA759 was lined up with taxiway C, and the PAL115 crew turned on that airplane’s landing lights to alert the ACA759 crew. Video information showed that PAL115’s landing lights illuminated the surface in front of that airplane as well as the tail and side of the first airplane on taxiway C, UAL1. About this time, the UAL1 captain made the first of two transmissions, on the tower frequency, about the position of an airplane above the airport surface; these transmissions were also available cues for the incident flight crew to recognize ACA759’s misalignment.107 In addition, about this time, ACA759 descended to an altitude at which its landing lights would have illuminated the environment below. The appearance of airplanes on the surface (especially given that the controller had just advised the flight crew that runway 28R was clear) and the lack of runway markings on the surface should have been additional cues indicating that the airplane was not aligned with runway 28R. During postincident interviews, the captain and the first officer were unable to identify a specific triggering factor in the environment that led to the decision to initiate and call for, respectively, a go-around. However, all of the cues mentioned above occurred within 6 seconds before the initiation of the go-around (2355:59 to 2356:05). That period is consistent with the time for pilots to recognize a cue, make a decision, and execute an action.

[ANALYSIS] When the captain aligned the airplane with taxiway C, he did not have any lateral guidance to indicate that the airplane was off course because of the first officer’s failure to tune the ILS frequency and the captain’s failure to verify that the approach was properly set up. Also, the first officer, as the monitoring pilot, was required to communicate all errors and situational awareness concerns to the captain. However, the first officer was heads down when the captain aligned the airplane with the taxiway because of the time that it took the first officer to locate the runway heading information (in response to the captain’s instruction to set the runway heading) and the time needed for the first officer to set the missed approach altitude and runway heading. As a result, neither crewmember realized, once the airplane passed the final waypoint, that the airplane was not aligned with the intended landing runway. The threat and error management model further indicated that, if the flight crew mismanaged the threat, then a procedural, communication, or handling error could occur. If such an error occurs and the crew traps the error, then safety margins will be maintained. Even though the tower controller confirmed that runway 28R was clear, the flight crew recognized that something was not right with the approach about the time that the airplane passed over the seawall. According to the captain, the first officer called for a go-around at the same time as the captain initiated the go-around maneuver, thereby preventing a collision on the taxiway. Even though the flight crew performed a go-around at that point, safety margins were severely reduced given ACA759’s proximity to the ground (below 100 ft) before the airplane began climbing and the minimal distance between ACA759 and the airplanes on taxiway C. Thus, a flight crew-induced aircraft state that compromised safety—the bottom level of the threat and error management model——resulted from ineffective threat and error management. The NTSB concludes that errors that the flight crewmembers made, including their false assumption that runway 28L was open, inadequate preparations for the approach, and delayed recognition that the airplane was not lined up with runway 28R, reflected breakdowns in CRM and led to minimal safety margins as the airplane overflew taxiway C. NTSB Aircraft Incident Report 47 2.3 Reasons for Flight Crew’s Misalignment With Taxiway C The flight crewmembers had recent experience flying into SFO at night: the captain reported that he had flown into SFO one or two times during the previous 4 months, and the first officer was the pilot flying on a flight to SFO 2 nights before the incident. The flight crewmembers’ training records indicated no issues with identifying airport surfaces, flying stabilized approaches, and flying visual approaches.

[ANALYSIS] Thus, such technology, if it had been installed on the incident airplane, could have helped the flight crew identify the surface misalignment and could have resulted in a go-around that was performed at a safer altitude (before the airplane was dangerously close to other airplanes), thereby improving safety margins. NTSB Aircraft Incident Report 57 FAA data (as of August 3, 2018) indicated that 85% of wrong surface landings involved general aviation airplanes.116 Thus, additional information about an airplane’s position relative to a runway on final approach would benefit all pilots of airplanes landing at primary airports within Class B and Class C airspace.117 The NTSB concludes that flight safety would be enhanced if airplanes landing at primary airports within Class B and Class C airspace were equipped with a cockpit system that provided flight crews with positional awareness information that is independent of, and dissimilar from, the current ILS backup capability for navigating to a runway. Although Honeywell’s SmartLanding system provides an alert if a potential taxiway landing is predicted, the NTSB understands (from conversations with the FAA) that there are current limitations on the widespread implementation of taxiway mapping.118 However, for a system that provides an alert when an airplane is not aligned with a runway surface, only the airplane position and the runway location are required, which are currently available in many transport-category airplanes. As a result, the NTSB believes that developing a system that provides an alert for non-runway surface alignment would be feasible in the near term with existing technology.119 Therefore, the NTSB recommends that the FAA establish a requirement for airplanes landing at primary airports within Class B and Class C airspace to be equipped with a system that alerts pilots when an airplane is not aligned with a runway surface. The system described in Safety Recommendation A-18-25 would alert if an airplane was predicted to land on a non-runway surface, such as a taxiway. The NTSB recognizes that the technology for a system that provides an alert to pilots when an airplane is not aligned with the intended runway surface may not currently be available. Such a system would also need to define and compare information about the intended runway (from ATC clearance instructions and/or the airplane’s navigation system) with the airplane’s runway alignment on final approach. 116 The FAA’s data also showed that 85% of wrong surface landings involved ATCT facilities that were considered to be between levels 4 and 9, indicating operations at small- or medium-sized ATCT facilities. These facilities might not have as many controllers and some of the equipment that level 10 through 12 ATCT facilities have. Thus, it is important for pilots operating at airports with level 4 through 9 ATCT facilities to have cockpit technology to aid in determining an airplane’s position relative to a runway on final approach. 117 According to the FAA, primary airports have more than 10,000 passenger boardings (enplanements) each year.

[ANALYSIS] However, the flight crew could not have verified that the airplane was tracking close to the extended runway 28R centerline given that the extended centerline information would not have been electronically depicted (because the ILS/localizer frequency had not been tuned). NTSB Aircraft Incident Report 44 captain requested that the first officer contact the tower controller to verify that the runway was clear. According to the ATC voice recording and the NTSB’s airplane performance study, when the flight crew transmitted, “we see some lights on the runway there, across the runway. Can you confirm we’re cleared to land?” the airplane was passing through an altitude of 300 ft. The controller had scanned the runway just before receiving the flight crew’s transmission. While receiving the transmission, the controller scanned the ASSC and radar displays to check for conflicts and scanned runway 28R again. The controller confirmed that the airplane was cleared to land and stated that no other airplanes were on the runway. At that point, the airplane was passing through an altitude of 200 ft. The airplane performance study for this incident showed that ACA759 continued the approach and flew over the first airplane on taxiway C (UAL1) at an altitude of 100 ft and that the flight crew initiated a go-around when ACA759 was at an altitude of 89 ft.92 The airplane performance study also showed that ACA759 flew over the second airplane on taxiway C (PAL115) at an altitude of 60 ft before the airplane began climbing, which resulted in only 10 to 20 ft of vertical separation between the ACA759 and PAL115 airplanes. The NTSB concludes that the flight crew-initiated, low-altitude go-around over the taxiway prevented a collision between the Air Canada airplane and one or more airplanes on the taxiway. Because of the lack of CVR data for the incident flight, the NTSB could not determine what information the pilots discussed, if anything, before and during the go-around. Nevertheless, the NTSB was able to determine the cues that likely led to the flight crew’s recognition of its alignment error and the crew’s initiation of the go-around maneuver, as discussed in section 2.3.2.2. The controller recalled that, when ACA759 was about 1/10 mile on short final, the airplane looked “extremely strange” regarding its location relative to runway 28R, taxiway C, and the airplanes along the taxiway. When the UAL1 captain stated, over the tower frequency, “where is that guy going?” at 2355:59, the controller expressed confusion regarding who had made the transmission.93 (The UAL1 captain did not identify himself during the transmission.) At that point, the controller, who had just checked runway 28R twice, was likely trying to process what he was seeing and hearing.

[ANALYSIS] These facilities might not have as many controllers and some of the equipment that level 10 through 12 ATCT facilities have. Thus, it is important for pilots operating at airports with level 4 through 9 ATCT facilities to have cockpit technology to aid in determining an airplane’s position relative to a runway on final approach. 117 According to the FAA, primary airports have more than 10,000 passenger boardings (enplanements) each year. Per the FAA’s Aeronautical Information Publication, Class B airspace is generally “surrounding the nation’s busiest airports in terms of IFR operations or passenger enplanements.” Class C airspace is generally “surrounding those airports that have an operational control tower, are serviced by a radar approach control, and that have a certain [minimum] number of IFR operations or passenger enplanements.” 118 The FAA indicated that one limitation is getting teams from (or contracted by) the FAA to every airport in the FAA’s database to survey taxiways, which would be a significant undertaking for a benefit that could be gained by installing a cockpit system that provides flight crews with positional awareness information by notifying them when an airplane is not aligned with a runway surface. 119 The NTSB is investigating the August 10, 2018, incident involving a Gulfstream IV airplane operated by Pegasus Elite Aviation as PEGJET flight 19 at Philadelphia International Airport, Philadelphia, Pennsylvania. The airplane, which was operating under 14 CFR Part 135 as a charter flight, was cleared to land on runway 35 about 2050 EDT. During the visual approach, the airplane aligned with taxiway E. About 0.1 mile from the end of taxiway E, the pilot initiated a go-around, and the airplane overflew four air carrier airplanes on taxiway E during the climb. The incident airplane came within about 200 ft of the first airplane on the taxiway. At the time of the approach, the runway 35 runway end identifier lights and the precision approach path indicator lights were out of service. The seven airplane occupants were not injured, and the airplane was not damaged. For more information, see NTSB incident number DCA18IA265. The circumstances of this incident demonstrate the importance of equipping airplanes with additional positional awareness technology when landing at primary airports in Class B and Class C airspace because of the risk to passengers in airplanes on a taxiway. NTSB Aircraft Incident Report 58 Honeywell’s systems are examples of existing technology that can accurately identify a runway surface when a pilot is on final approach.120 However, the NTSB is not aware of an automated system that has reliably demonstrated the ability to indicate whether an airplane is aligned with the specific runway for which it has been cleared. Such a system would further improve safety, especially at airports with parallel runways (which, according to the FAA’s wrong surface landing video, account for 75% of wrong surface landings), and provide a longer term and more robust solution to wrong surface landings.

[ANALYSIS] The NTSB concludes that, if an airplane were to align with a taxiway, an automated ASDE alert would assist controllers in identifying and preventing a potential taxiway landing as well as a potential collision with aircraft, vehicles, or objects that are positioned along taxiways. The taxiway arrival prediction capability that was implemented at SEA (and was scheduled to be evaluated and implemented at other feasible ASDE-X locations) could be expanded to other ASDE system models (ASDE-3 and ASSC). Therefore, the NTSB recommends that the FAA modify ASDE systems (ASDE-3, ASDE-X, and ASSC) at those locations where the system could detect potential taxiway landings and provide alerts to air traffic controllers about potential collision risks. 2.6 Runway Closure Markers A runway closure marker with a lighted flashing white “X” appeared at the approach and departure ends of runway 28L when it was closed, including on the night of the incident. The lighted “X” was consistent with the specifications in FAA AC 150/5345-55A, “Specification for L-893, Lighted Visual Aid to Indicate Temporary Runway Closure.” Although the runway closure marker might have been effective at preventing a takeoff from or a landing on runway 28L when it was closed (the specific risks that the lighted “X” was designed to address), the runway closure marker did not capture the attention of the incident flight crew as the airplane approached the airport while aligned with taxiway C. The lighted “X” runway closure marker was not designed to address the possibility that a flight crew could misidentify a runway surface due to ineffective signaling of a runway closure. Although air traffic controllers can provide instructions to pilots about the closure of a runway, NOTAMs and ATIS broadcasts/messages are the primary means to inform pilots about runway closures. However, the information about runway closures provided in NOTAMs and ATIS broadcasts/messages is not necessarily a reliable means for ensuring that pilots are aware of the NTSB Aircraft Incident Report 64 closure information.126 As previously stated, although the runway 28L closure on the night of the incident was indicated in NOTAM and ATIS information that the incident flight crew received, that information was not effective in preventing the flight crew from misaligning the airplane during the approach to runway 28R.127 The incident flight crew and the flight crew of DAL521 (which landed on runway 28R about 4 minutes before the incident occurred) stated that they did not see a lighted “X” on runway 28L to indicate that it was closed. Given the current specifications for the runway closure marker, these flight crews likely did not see the “X” because their airplanes were not aligned with runway 28L. This situation highlights the need for a surface-based system with conspicuous and unambiguous visual cues that clearly indicate when a runway is closed, even if an approaching airplane is not aligned with the closed runway.

[ANALYSIS] The NTSB found multiple reports of altitude and lateral deviations in which the pilot cited the formatting of information as a related issue, which further indicates that information display improvements are needed for text-based resources. For more information, see exemplar cases (ACNs [accession numbers] 1282309, 1409095, 1414185, 1447318, 1540033, 1544364, and 820437) in the ASRS database (https://asrs.arc.nasa.gov/search/database.html, accessed September 14, 2018). 100 The second and third airplanes on taxiway C, an Airbus A340 and a Boeing 787, respectively, had wing spans of about 200 ft, which was also the runway 28R width. (The first airplane on taxiway C, a Boeing 787, also had a wing span of 200 ft, but the airplane was oriented perpendicularly to the taxiway. Thus, those wingtip lights would not have created the illusion to the incident flight crewmembers that they had aligned ACA759 with runway 28R.) NTSB Aircraft Incident Report 50 been consistent with features associated with approach lighting. Another cue that would have supported the crew’s perception was the presence of runway and approach lights on runway 28R, which would also have been present on runway 28L when open. However, the runway and approach lights on runway 28L were off, and the construction lighting that was reported on the runway 28L surface had features that were consistent with ramp lighting. A psychological concept associated with perception and decision-making that can allow a mistaken assessment to persist is expectation bias, which refers to the manipulation of perceived elements to values consistent with a person’s expectation (Bhattacherjee 2001). A similar concept, confirmation bias, results from a tendency to primarily seek out confirming evidence of a belief while spending less effort seeking out negative evidence that can disconfirm the belief (Nickerson 1998). Thus, expectation bias and confirmation bias can cause a person’s incorrect belief to persist despite available contradictory evidence. Both biases occur as part of basic information processing, and a person may not be actively aware of such biases at the perceptual level. In this report, the term “expectation bias” also describes the effects of confirmation bias. Expectation bias is not a new phenomenon in aviation. The NTSB investigated numerous accidents and incidents that involved pilot errors resulting from expectation bias, particularly in night VMC when fewer cues were available to pilots to aid in airport and runway identification. For example, in January 2014, a Boeing 737 landed at the wrong airport in Branson, Missouri, in night VMC. The flight crew expected that the visually identified airport and runway were the intended destination and did not reference cockpit displays to verify the airport and runway. As a result, the airplane landed on runway 12 at M. Graham Clark Downtown Airport instead of runway 14 at Branson Airport.

[ANALYSIS] When the UAL1 captain stated, over the tower frequency, “where is that guy going?” at 2355:59, the controller expressed confusion regarding who had made the transmission.93 (The UAL1 captain did not identify himself during the transmission.) At that point, the controller, who had just checked runway 28R twice, was likely trying to process what he was seeing and hearing. The controller stated, during a postincident interview, that the transmission seemed “out of context.” The controller had no reason to think that ACA759 was lined up with taxiway C before he observed the airplane looking “extremely strange.” The ACA759 flight crew had reported that the airplane was on approach to runway 28R, and the ASSC display had previously predicted that the airplane would be landing on runway 28R. Also, the controller had not previously seen an airplane align with taxiway C. Further, the distance and angle (parallax) of the tower cab relative to the approach end of runway 28R and taxiway C would have made it difficult for the controller to visually recognize that ACA759 was aligned with the taxiway instead of the runway, especially at night and with the lights from the construction on runway 28L and airport vehicle movements. 92 As previously stated, all altitudes in this report are agl (unless otherwise indicated). 93 The ATC voice recording indicated that, 2 seconds after the UAL1 captain stated, “where is that guy going?” and 2 seconds before he stated, “he’s on the taxiway,” the controller stated to himself, “who is…talk?” NTSB Aircraft Incident Report 45 When the UAL1 captain stated, 4 seconds after his first transmission, “he’s on the taxiway,” the controller might have realized what had happened with ACA759. Because the flight crew had already begun the go-around maneuver (at 2356:05), the airplane was climbing at the time of the controller’s go-around instruction (2356:09). The controller subsequently told the flight crew that the airplane appeared to have been lined up with taxiway C. The NTSB concludes that the controller responded appropriately once he became aware of the potential conflict. A factor that precluded the controller from determining sooner that a potential conflict existed was the ASSC system’s lack of capability to detect a taxiway landing and provide an alert, as discussed further in section 2.5.2. 2.2.5 Crew Resource Management Breakdown Air Canada’s CRM manual for pilots and the company’s CRM competency guide addressed, among other subjects, situational awareness, workload management, active monitoring, and threat and error management. Several crew actions/inactions during the incident flight demonstrated breakdowns in CRM, many of which were manifested as noncompliance with Air Canada’s standard operating procedures.

[ANALYSIS] Also, the captain stated that he had never seen runway 28L “dark” and that none of his previous landings at SFO occurred when a runway was closed. A runway closure marker with a flashing white “X” was placed at the threshold of runway 28L to indicate the runway closure. However, the flashing “X” would not have been in the flight crew’s direct field of view because the “X” was oriented toward the runway 28L final 95 Industry analysis of traffic patterns at SFO showed that airplanes landing on runway 28R tended to stay more to the right of course when runway 28L was open than when it was closed. NTSB Aircraft Incident Report 48 approach corridor and the airplane was not aligned with runway 28L.96 Also, the flash rate might have been too slow to capture the crew’s attention.97 (The conspicuity of the “X” to flight crews on final approach is further discussed in section 2.6.) The NTSB concludes that the flight crewmembers’ lack of awareness about the runway 28L closure and the crewmembers’ previous experience seeing two parallel runways at SFO led to their expectation to identify two runway surfaces during the approach and resulted in their incorrect identification of taxiway C instead of runway 28R as the intended landing runway. The NTSB considered the presentation and priority of the runway 28L closure information compared with other information that the flight crew received. The flight release package was 27 pages long and consisted of, among other items, route, weather, and NOTAM information. The NOTAM indicating the runway 28L closure (“RWY 10R/28L CLSD”) appeared on page 8 of the package, which was also the second page of NOTAM information, under the gray highlighted heading “DESTINATION” (which appeared on the previous page). Features of the NOTAM text emphasized the closure information, such as the use of bold font for the words “RWY” and “CLSD” and a “**NEW**” designation in red font with asterisks before the NOTAM text, as shown in figure 10.98 However, this level of emphasis was not effective in prompting the flight crewmembers to review and/or retain this information, especially given the NOTAM’s location (toward the middle of the release), which was not optimal for information recall. A phenomenon known as “serial position effect” describes the tendency to recall the first and last items in a series better than the middle items (Colman 2006). The ACARS message providing ATIS information Quebec, as displayed in the cockpit, was 14 continuous lines with all text capitalized in the same font. As shown in figure 11, the NOTAM indicating the runway 28L closure appeared at the end of line 8 and the beginning of line 9. The uniform presentation of the ATIS information could have contributed to the flight crew’s oversight of the runway closure information.

[ANALYSIS] The flight crew was properly certificated and qualified in accordance with Canadian regulations, Air Canada requirements, and 14 CFR Part 129. Flight crew medical conditions. The flight crew held valid and current Canadian medical certificates. The captain and the first officer reported no medications, medical conditions, or sleep disorders during their required medical examinations and during postincident interviews. Airplane mechanical conditions. The airplane was properly certificated, equipped, and maintained in accordance with Canadian regulations and 14 CFR Part 129. No evidence indicated any structural, engine, or system failures. Airport lighting. Runway 28R (the intended landing runway for the flight) and parallel taxiway C at SFO met the requirements for lighting in 14 CFR Part 139. The runway 28R approach lights were on as the incident airplane approached SFO. Title 14 CFR 139.311 required taxiways to have one of four types of lighting systems, one of which was centerline lights, which were present and illuminated on taxiway C. The runway and taxiway centerline lights were set at step 1 (out of 5), which was appropriate for the weather conditions on the night of the incident. NTSB Aircraft Incident Report 40 Thus, the NTSB concludes that none of the following were factors in this incident: (1) flight crew qualifications, which were in accordance with Canadian and US regulations; (2) flight crew medical conditions; (3) airplane mechanical conditions; and (4) airport lighting, which met US regulations. 2.2 Incident Sequence 2.2.1 Notification of Runway 28L Status SFO runway 28L was scheduled to close at 2300 on the night of the incident due to construction, and a NOTAM in the flight release for ACA759 provided information about the closure. The captain stated that he and the first officer discussed the runway 28L closure while at the departure gate but that they did not place much emphasis on this information because the captain thought that the flight would land before the runway would be closed.84 (The NTSB notes that the flight was originally scheduled to land 3 minutes after the runway 28L closure.) However, the airplane pushed back from the gate 30 minutes late due to the delayed arrival at YYZ of the airplane to be used for the flight and took off about 49 minutes later than the scheduled takeoff time. Even though the flight crewmembers knew, during their preflight preparations, about ACA759’s delayed departure, no evidence indicated that the crewmembers reconsidered the importance of the NOTAM information at that time or as the airplane approached SFO.

[ANALYSIS] The NTSB classified Safety Recommendation A-03-5 “Closed—Acceptable Action” on January 30, 2012. 127 The NTSB has investigated other events in which pertinent NOTAM information was missed; for example, see NTSB accident numbers DFW07CA092, CEN12LA229, and CEN14FA505. NTSB Aircraft Incident Report 65 technology have occurred. These advances could allow more conspicuous attentional capture features to be incorporated into a runway closure marker’s design. Such features, which include varying the flash pattern, incorporating strobe lights, and/or creating apparent movement, might direct a pilot’s attention to a closed runway better than the current design of the lighted “X.” The NTSB concludes that increased conspicuity of runway closure markers, especially those used in parallel runway configurations, could help prevent runway misidentification by flight crews while on approach to an airport. Therefore, the NTSB recommends that the FAA (1) conduct human factors research to determine how to make a closed runway more conspicuous to pilots when at least one parallel runway remains in use and (2) implement a method to more effectively signal a runway closure to pilots during ground and flight operations at night. NTSB Aircraft Incident Report 66 3. Conclusions 3.1 Findings 1. None of the following were factors in this incident: (1) flight crew qualifications, which were in accordance with Canadian and US regulations; (2) flight crew medical conditions; (3) airplane mechanical conditions; and (4) airport lighting, which met US regulations. 2. The first officer did not comply with Air Canada’s procedures to tune the instrument landing system (ILS) frequency for the visual approach, and the captain did not comply with company procedures to verify the ILS frequency and identifier for the approach, so the crewmembers could not take advantage of the ILS’s lateral guidance capability to help ensure proper surface alignment. 3. The flight crew’s failure to manually tune the instrument landing system (ILS) frequency for the approach occurred because (1) the Flight Management System Bridge visual approach was the only approach in Air Canada’s Airbus A320 database that required manual tuning of a navigation frequency, so the manual tuning of the ILS frequency was not a usual procedure for the crew, and (2) the instruction on the approach chart to manually tune the ILS frequency was not conspicuous during the crew’s review of the chart. 4. The first officer’s focus on tasks inside the cockpit after the airplane passed the final waypoint reduced his opportunity to effectively monitor the approach and recognize that the airplane was not aligned with the intended landing runway. 5. The flight crew-initiated, low-altitude go-around over the taxiway prevented a collision between the Air Canada airplane and one or more airplanes on the taxiway. 6. The controller responded appropriately once he became aware of the potential conflict. 7. Errors that the flight crewmembers made, including their false assumption that runway 28L was open, inadequate preparations for the approach, and delayed recognition that the airplane was not lined up with runway 28R, reflected breakdowns in crew resource management and led to minimal safety margins as the airplane overflew taxiway C. 8. The flight crewmembers’ lack of awareness about the runway 28L closure and the crewmembers’ previous experience seeing two parallel runways at San Francisco International Airport led to their expectation to identify two runway surfaces during the approach and resulted in their incorrect identification of taxiway C instead of runway 28R as the intended landing runway.

[ANALYSIS] Therefore, the Safety Board reiterates its concerns exoressed in Safety Recommendation A-89-65 and urges the FAA to consider amending its policies for evaluating the closeout of NASIP findings. 2.8 Analysis of Other Corrective Actions 2.8.1 The FAA’s Runway Incursion Prevention Plan This accident, an earlier collision in Atlanta, and a later collision in ios Angetes, spurred the FAA into updating and finalizing its runway incursion prevention efforts, although work in this area was initiated several years ago. The FAA’s Runway Incursion Prevention Plan appears to be thorough and is now under a single manager. High technology systems such as the Airport Movement Area Safety System (AMASS), Advanced Airport Surface Detection Equipment (ASDE-3), and Airport Surface Traffic Automation (ASTA), when perfected, should enhance the safety of airport ground operations considerably. The concept of the formation of "demonstration airports" to exhibit and test new or different devices and surface marking methods is valid, and the selection of the four specific demonstration airports was done in an appropriate manner. The Safety Board is also aware of other new technologies, such as the satellite-based Globa) Positioning System, that could be included in future runway incursion prevention systems. 2.8.2 Detroit Metropolitan/Wayne County Airport Since the accident, Detroit Metropolitan/Wayne County Airport personnel corrected a number of airfield discrepancies that were discovered during the Safety Bcard’s investigation. The rheostat switches on the control tower airfield lighting panel were replaced with switches equipped with stops in accordance with AC 156/5345-3D. Reflective paint is now being used for all airfield markings, including taxiway centerlines and runway hold lines. All faded taxiway centerlines identified as faded during the investigation have been repainted, and a program to repaint markings when they are discovered faced is in effect. A purchase contract for a replacement taxiway hold position tight was awarded and the light is being fabricated. Semiflush runway edae lights in the runway 3€/2)c-9/27 intersection area are scheduled to be installed by September 1991. An experimental system of outlining taxiway centerline markings on concrete areas of the taxiways in black paint to improve contrast is in effect. Lastly, permanent removal of the Outer 4 taxiway between the Outer taxiway and the runways is also scheduled to occur in September 1991. 74 2.8.3 Northwest Airlines, Inc. The new seven-item low-visibility taxi section that NWA is incorporating into its Flight Operations Manual is an excellent tool to inform its pilots of the dangers of aircraft movement in instrument weather conditions.

[ANALYSIS] At 1344:35, for the third time, he told the first officer to call the tower for assistance saying, "Well, tell] him we’re out here. We're stuck." The first officer still did not comply, but he did respond inaccurately (again) with, "That’s 09.” At 1344:47, the captain finally asserted his authority. After two unsuccessful attempts on some unknown frequency or on interphone, he succeeded in informing the ground controller that they were on an funidentified] runway. Less than 1 minute prior to the collision, the captain exercised his command responsibility. 8y 1345:14, the first officer was apparently convinced that they were not only on a runway but that it was the active runway and so informed the captain. The captain relaved this information to the ground contro:ler at 1345:17. It was not unti? ~°~.33, 7 seconds prior to the collision, that the ground controller ordered fiight 1482 off the active runway. When the captain transmitted, “Yeah, it looks like we’re on 21 Center here," at 1345:17, he was asked to confirm this statement by the ground controlier. The captain then stated, "I believe we are, we’re not sure." Following the accident, the captain said that if he had been positive that he was on an active runway and that another airplane was bearing down upon him, he would have taxied off the runway onto the grass. In this instance, he was sufficiently aware that something was wrong that he intentionally taxied to the edge of the paved surface of the runway. In a previous accident investigation report (NTSB/AAR-84-10) concerning a runway incursion and subsequent collision between a Korean Airlines’ DC-10 and a Southcentral Air Piper PA-31 in 1983, the Safety Board addressed problems similar to the role reversal in the cockpit of the DC-9. That report stated: The captain’s statement indicates that he felt that the first officer, who had a higher level of recent experience at the airport than the captain, was more certain about the aircraft's location than the captain was.... The Safety Board believes that the first officer’s strong belief about their jocaticn may have influenced the captain’s decision to commence takeoff. The first officer’s confidence regarding being on the correct runway in the face of the captain’s uncertainties constituted a slight role reversal in that the captain’s overall command authority when deciding to take off was influenced by the first officer’s comments. In the past, the Safety Board has encouraged assertiveness training for first officers to exercise their responsibilities as part of the cockpit team; however, a companion responsibility for captains to exercise positive cockpit crew management must exist. 59 As a result of the investigation of the accident at DTW on August 16, 1987, involving NWA, the Safety Board issued Safety Recommendation A-88-71 to al] Part 121 air carriers, including NWA.

[CONCLUSIONS > FINDINGS] 3.1 Findings 1. All flight crewmembers, flight attendants, and air traffic controllers were properly certificated to perform their duties. to Visibilities at the tine and area of tre collision varied, with the lowest estimated horizontal visibility near 100 feet. The officiai prevailing visibility, as determined by National Weather Sarvice and Federal Aviation Administration personnel, was 1/4 mile. 3. The B-727 captain attempted a takeoff in runway visibility of less than 1/4 mile. 4. The runway centerline lights on runway 3C/21C were not iltuminated at the time of the accident. an fhe placement of taxiway signs, the conspicuity of taxiway ma;-kings, and runway lighting were inadequate at DTW at the time of the accident. 6. The DC-S flightcrew failed to follow their assigned routing in the taxiway Oscar-6 area. 7. The flightcrew contributed to their confusion by failing to taxi toward and intersect the centerline of the Inner taxiway where it paralleled the edge of the concrete as they left the 10, il. 12. 13. 14. 15. 16. 17. 77 parking area. If they had done so, the centerline leading to Oscar-6 would have been more apparent to them. The complex intersection of taxiviay Oscar-4, and runways 09/27 and 3C/21C was a recognized danger area with a strong potentia! for runway incursions but was nevertheless inadequately marked. The pilots of the DC-9 failed to consistently cross-check the airptane’s heading with the headings of their taxi rouviag. A reversa}}] of command roles accurred during the accident sequence in which the first officer made most of the decisions regarding taxi activity and the captain tacitly relinquished his command role. The first officer misled the captain concerning his familiarity with DTW and failed to follow the capiain’s direct instructions on three occasions prior to the runway incursion. If the captain and first officer of the DBC-9 had received thorough training in cockpit resource management, the comnand role reversal might not have occurred. The captain of the DC-9 questioned his pcsition a full 53 seconds before the collision; yowever, neither he nor the first officer advised the ground controller of their uncertainty at that time. If they had done so, the loca? centrolier might have taken action to prevent the 8-727 takeoff. The east ground controller missed several opportunities to take appropriate action to resolve confusion on the part of the DC-9 crew. The east ground controller, after he realized that the DC-9 might have taxied onto an active runway, did not take timely action to correct the problem.

[ANALYSIS] Immediately after this comment, the first officer stated, “Yeah, this is [runway] 9. Were, we’re facing 160 [degrees], yeah. Cleared to cross it." The only taxiway segment in the Oscar 4 area having a heading of 1609 leads directly to the intersection of runway 9/27 and the active runway 3C/21C. However, neither the captain nor the first officer noticed this fact. The captain, his doubts apparently somewhat eased by the first officer’s confidence, then asked, "We’re cleared to cross?" The first officer replied confidently, "Yeah, we’re cleared ta cross.” The captain then asked, “Which way do I go? Right?" The first officer responded, "Yeah." This conversation was representative of the entire taxi sequence--the role reversal in the cockpit of the DC-9. The captain was about to complete a direct order to the first officer to make a radio call to the tower concerning their predicament. The first officer instead interjected his statement that they were on runway 9. The captain believed him and resumed a subordinate role when he asked the first officer more questions as he taxied the airplane southeasterly toward the active runway. At 1342:56, the captain evidently began to have real doubts ab -ut their location when he stated, "This, this is the active runway here isn’t 1t?" The first officer, perhaps by then less confident of his navigation, stated, "This is, should be 9 and 27. It is. Yean, this is 9/27." The Safety Beard believes that about this time, 1343:08, the airplane first entered the active runway, although it had crossed the hoid line for the runway earlier. Shortly thereafter, the captain apparently saw white lines that convinced him that they were not on a taxiway. He stopped the airplane, setting the parking brake. At 1343:35, he gave a complete order to the first officer to, "Give him a cal}} and tell him that, ah, we can’t see nothin’ out here." The first officer did not comply with this order and, after a lapse of about 13 seconds, responded incorrectly to another ground control request for their position. The Safety Board believes that if the first officer had obeyed the captain immediately, the air traffic controllers might have taken more timely action to step the B-727/ takeoff. According to “he captain’s testimony and 58 the CVR transcript, he then released the parking brake and began to angle off to the left of the runway as he began ta have more doubts about their location. At 1344:35, for the third time, he told the first officer to call the tower for assistance saying, "Well, tell] him we’re out here. We're stuck." The first officer still did not comply, but he did respond inaccurately (again) with, "That’s 09.” At 1344:47, the captain finally asserted his authority.

[ANALYSIS] The Board believes, however, that when the off-duty controller asked the iocal controtier if he was going to change the official visibility, the local controller should not have arbitrarily dismissed her query. 61 2.3.2 Ground Controller’s Actions In analyzing this accident in retrespect, the Safety Board examined the actions that could have been taken by the ground controller to prevent the runway incursion. After determining that the DC-9 had missed Oscar 6 and was in the vicinity of Oscar 5 after having inadvertently turned eastbound on the Outer taxiway, the controller had some options. First, because the Oscar 4 area had been identified as a potentia? runway incursion hazard in materials available to him, the controller could have kept the airplane away from that area by directing it back to the Oscar 6 throat via Oscar 5 and the Inner taxiway. The Safety Board does not believe that many controllers would have used this option, particularly when communicating with a professiona: airline crew presumably familiar with their *:b airport. Having opted to route the flight toward the Oscar 4 arez nowever, the controller could have taken other precautions. He could have begun issuing progressive taxi instructions, informing the crew to continue to the next taxiway intersection--identifiable by the sign for Outer/Xray--and hold short. Furthermore, recognizing the low-visibility conditions and the problems already experienced by the DC-9 crew, he could have requested the local controller to suspend takeoff activity until he was certain that the DC-9 was in fact across runway 9/27 clear of the Oscar 4 area and established on taxiway Xray. In any event, the controllers clearance "continue to Oscar 4 and then turn right on “ray” was not precise because the airplane would not actually intersect the centerline of the Oscar 4 taxiway, nor would the DC-9 crew see any signs for Oscar 4, when negotiating the acute right turn onto taxiway Xray. The Safety Board does not believe however that the actual clearance should have confused the flightcrew since the designation Oscar 4 on the airport diagram available to the crew appears to encompass the intersection of the Outer taxiway and Xray. Although the Safety Board believes that the ground controller could have selected a more conservative taxi routing, control technique, and clearance phraseology, it does not believe that his actions were deficient until he became aware that the flightcrew was encountering difficulty in the Oscar 4 area. This awareness occurred at 1345:02 when the captain of the DC-9 admitted an uncertainty atout his position. This was 47 seconds after the B-727 was cleared for takeoff and only one second before the sound of increasing engine noise was audible on the B-727’s CVR. The Safety Board recognizes that minimum time was available for controllers to act to prevent the accident. Nonetheless, the Board believes that the ground controller should have informed the local controller and his area supervisor immediately that he was unsure of the DC-9’s position.

[ANALYSIS] For instance, the Oscar 6 sign at the intersection of Oscar 6 and the Outer taxiway misled the flightcrew into believing that they were on Oscar 6 when they were not. Adding an arrow and an OTR/arrow to this sign might clarify its meaning. Along the Outer taxiway, there were no signs to indicate to the pilots that they were approaching the Oscar 4 taxiway. It is logical ta assume that Oscar 4 would be the next available taxiway after Oscar 5, when taxiing east, but in this case, the turnoff to Xray taxiway is next. In fact, several investigators, some of whom were current airline pifots, were confused by the signage in this area when they observed it on a clear day 66 after the accident. The inspectors of the signage from the airport and the FAA are not airline pilot. and, in some cases, are not pilots of any type of aircraft. The Safety Board believes that more user input should have been sought when the decision was made to place some signs at DTW. It recommends therefore that a survey be conducted of DTW signage for the purpose of developing signage that is more understandable to line pilots. Input from Tine pilots, rather than management or instructor pilots, should be a vital part of this survey. Also, the two hold lines in the Oscar 4 area were pzerallel to runways instead of perpendicular to their respective taxiways. F' ightcrews expect hoid lines to be at right angles to taxiway centerlines and, in this accident, the DC-9 crew may have seen the yellow markings but cnuld have failed to recognize them as hold lines because of the angle relative to the taxi path. The absence of runway edge Jights on runway 3C/21C in the Oscar 4/runway intersection area also probably contributed to the fligttcrew’s actions. If the lights had been imbedded in the pavement at intervals of 200 feet, as recommended by the AC, the DC-9 pilots would probably have noticed chem before the runway incursion and stopped taxiing. The Safety Board notes that the single runway edge light that the captain eventually observed prompted him to taxi to the left of the runway centerline during the incursion. Although it was not a direct factor in this accident, the Safety Board discovered that the centerline Tights on 3C/2Z1C were not annotated on the National Ocean Service or Jeppesen airport diagrams. This is an FAA responsibility and the FAA inspectors responsible for DTW should have ensured that the diagrams were accurate. These diagrams are used by pilots to predict what they will see when they taxi out fer departure and takeoff. It is also important for pilots te be aware of the runway lighting configuration white they arz conducting instrument approaches. The lighting panel in the tower is an airport responsibility. The Safety Board believes that the tower controllers thought the centerline yights were on because the rheostat for the lights was at or near the step 5 {{highest}} setting.

[ANALYSIS] The B-727 captain believed that since the ATIS stated that the visibility was 1/4 mile and he had “adequate visual references," he was Tegal in attempting the takeoff. His concept of adequate visual references 65 was the ability to maintain the runway centerline during the takeoff run, which he did during the collision with the DC-9 and during the aborted takeoff. He also stated that since he was not a trained weather observer, he could not be expected to question the ATIS information and had to accept the 1/4 mile ATIS observation as valid. The Safety Board believes it true that a pilot might have difficulty determining the visibility to within 300 or 400 feet. In this case, however, it was apparently obvious that the visibility was far less than 1,300 feet or 1/4 mile. 2.5 DTW Signage, Lighting and Markings The Safety Board recognizes that maintenance of al] signs, lights and pavement markings on an airport as large as DTW is a demanding task. However, some rather obvious shortcomings in this area were apparent. Although most of these shortcomings are not violations of any FARs, they reflect a disregard fcr the guidelines in several FAA advisory circulars concerning airport operations. The FAA was aware of some of these shortcomings and could have taken actions to correct them prior te the accident. The investigation revealed several areas of faded or nearly invisible taxi lines on the airfield, especially near the area where the DC-9 was taxiing. These deficiencies may have been a factor in the DC-9 flightcrew’s incorrect decision to turn ieft onto the Outer taxiway. However, photos taken after the accident showed that the yellow lines leading to Oscar 6 were clearly visible from the centerline of the Inner taxiway where it paralleled the edge of the ramp near the fire house. ‘Thus, if the flightcrew had acquired the centerline of the Inner taxiway as it paralleled the edge of the ramp near the fire house, the fork between Oscar 6 and the easterly heading portion of the Outer taxiway would have been more evident to them. The Safety Board believes that the repainting of the faded taxiway centerlines shoutd be performed as soon as they are noted during daily airport inspecticns instead of during a set schedule for overall airport restriping. Another confusing factor was the Oscar 6 sign located on the isiand between the Inner and Outer taxiways. Although the investigation determined that the size, coloration, and lighting of the airport signs in question met or exceeded regulatory requirements, the location and annotation of several signs observed by the DC-9 crew bear further discussion. For instance, the Oscar 6 sign at the intersection of Oscar 6 and the Outer taxiway misled the flightcrew into believing that they were on Oscar 6 when they were not. Adding an arrow and an OTR/arrow to this sign might clarify its meaning. Along the Outer taxiway, there were no signs to indicate to the pilots that they were approaching the Oscar 4 taxiway.

[ANALYSIS] Similarly, procedures requiring the use of progressive taxi during low-visibility conditions could provide more contro] and awareness to ground controllers of aircraft locations on the airport. The implementation of procedural redundancies could involve general national guidelines for supervision, as well as site-specific guidelines and procedures for certain airports with unique operating envirenments. Therefore, the Safety Board believes that the FA should immediately develop and implement procedures and policies to provide human redundancy of critica] controlier tasks, and should expedite the development and installation of redundant hardware systems. 2.4 B-727 Captain’s Decision to Take Off Considerable evidence suggests that the visibility at the departure end of runway 3C was much less than 1/4 mile. For example, a Mesaba Airlines’ captain who was No. 2 for takeoff behind the B-727 testified that re was unable to see more than 100 feet while crossing runway 9/27. Moreover, they could barely see the first visual approach slope indicator light box 750 feet down the runway and about 600 feet from their position as they held at the departure end of runway 3C. Alsc, most significantly, the first officer on the B-727 announced that they did not have 1/4 mile visibility as they taxied onto the runway and applied takeoff power. The captain later stated that the first officer retracted that observation shortly thereafter, but the retraction was not recorded on the CVR. The first officer’s statement at 1345:08, "Definitely not a quarter mile but, ah, at Teast they’re calling it," and a tack of response by the captain, indicates two things. First, a lack of CRM, in that the captain did not respond to the first officer’s concern about the visibility in any manner, positively or negatively. Second, the last part of the first officer’s statement indicates a reliance, at least in this pilot’s mind, upon only the control tower for takeoff visibility information. In other words, the first officer appeared to believe that the takeoff was permissible as Tong as the contro! tower stated that the prevailing visibility was 1/4 mile. If the captain of the B-727 had decided not to take off on runway 3C because cr low visibility, his flight and others, including the DC-9, would have been directed to use one of the outer runways at DTW for departure. NWA’s takeoff minimum visibility for those runways, because of their enhancea lighting and visibility measuring equipment, was 600 feet runway visuai range. The B-727 captain believed that since the ATIS stated that the visibility was 1/4 mile and he had “adequate visual references," he was Tegal in attempting the takeoff. His concept of adequate visual references 65 was the ability to maintain the runway centerline during the takeoff run, which he did during the collision with the DC-9 and during the aborted takeoff.

[ANALYSIS] Specifically, the LC2 stated that she did not hear the flightcrew of SKW5569 state that they were at taxiway 45. If the flightcrew 64 of SKW5569 had stated, "we are at the taxiway 45 intersection, ready for takeoff," it is possible that the misidentification might not have occurred. The use of nonstandard words and conversational phraseology precipitates misunderstanding between pilots and controllers. The Safety Board’s Special Investigative Report entitled "Runway Incursions at Controlled Airports in the United States" (NTSB/SIR-86/01) disclosed that many runway incursions were attributable to the improper use of phraseology that resutted in miscommunications by controllers and pilots. The joint FAA/industry partnership to improve pilot/controller communication that produced the document "Call to Action,” published in 1988, provided further evidence that the most common and troublesome problem evident in the ATC system was the improper use of established and recommended phraseology by pilots and controllers. Neither the AIM nor the Air Traffic Control Handbook (7110.65F) contain specific phraseology to be used by pilots when requesting an intersection departure and by ATC personnel when issuing a position-and-hold clearance for an intersection departure. The Los Angeles accident provides vivid evidence that position-and-hold operations at intersecting points along runways continue to play a significant role in the runway incursion problem. The Safety Board believes that a solution to reducing misunderstandings and/or loss of situational awareness by pilots and controllers concerning intersection takeoffs is to establish clear and concise standard terminology for pilots and controllers. For example, pilot request: "Cessna N12345 request intersection takeoff from runway 24 Left at taxiway 45;" controller reply: "Cessna N12345, taxi into position and hold runway 24 Left at intersection 45." Recommended communication phraseology regarding the request for intersection departures should be incorporated into the appropriate section of the AIM. In addition, standard air traffic phraseology and procedures regarding position and hold at intersections should be incorporated into the Air Traffic Control Handbook (7110.65F). Moreover, the Safety Board believes that all pilots, general aviation and commercial, should be made aware of the events leading up to this accident through operations bulletins and safety seminars, such as the "Wings Pilot Proficiency Program." 2.6 Survival Factors The emergency response for this accident was timely and effective. The close proximity of Fire Station 80 to the accident site, coupled with the rapid response of ARFF units, facilitated personnel efforts to apply extinguishing agent to the external fires and to assist some of the passengers in egressing from the B-737. The Safety Board believes that these factors reduced injuries and saved lives. The Safety Board also found that the rapid availability of adequate numbers of ARFF-trained fire fighters, from both Fire Station 80 and off-airport structural fire companies, allowed ARFF personnel to implement an interior fire attack immediately.

[ANALYSIS] The Safety Board believes that a continuous access path of no less than 20 inches, as demonstrated by tests, is preferable to removing the seat adjacent to the exit or removing the seat and having a 20-inch or less access path. Furthermore, the Safety Board believes that the proposed compliance requirement of 6 months is necessary and reasonable because operators have had ample time to prepare for this proposed regulation. The Safety Board supports this rule and encourages the FAA to develop and issue a final rule at the earliest possible date. 2.7 Efforts to Reduce Runway Incursions The Safety Board has long been concerned about the runway incursion/ground collision issue. Based on that concern, the Board included this issue when it adopted the "Most Wanted" Safety Recommendations program in 1990. The issue continues to be a part of the "Most Wanted" list. This concern was heightened by two recent fatal accidents that preceded this accident. These previous accidents were the collision in Detroit, Michigan, on December 3, 1990, between Northwest Airlines flights 299 and 1482'> and ISunorthwest Airlines, Inc., Flights 1482 and 299, Runway Incursion and Collision, Detroit Metropolitan/Wayne County Airport, Romulus, Michigan, December 3, 1990" (NTSB/AAR-91/05) 71 the collision in Atlanta, Georgia, on January 18, 1990, between Eastern Airlines flight 111 and an Epps Air Service King Air Al100.1° The runway collision of USA1493 and SKW5569 involved controllerrelated factors identified in previous Safety Board reports. These factors are related to human performance and are being addressed in a number of different actions, including FAA and industry efforts to increase awareness of the nature and magnitude of the human performance problem, improved training and technological solutions that may reduce the workload, and a fail-safe redundancy for the human performance of air traffic controllers. The Safety Board is aware of several advanced concepts in airport surface traffic detection and automation that, when perfected and coupled with the correct match of hardware and location-specific software, could provide warnings to preclude accidents similar to the collision of USA1493 and SKW5569. For example, the FAA is currently testing an Airport Movement Area Safety System (AMASS). The AMASS system will use the data available in Airport Surface Detection Equipment (ASDE-3) and the Automated Radar Terminal System (ARTS) to identify potential incursions and will alert the controller so that timely corrective actions can be taken. The Safety Board fully supports the early development and installation of such systems at appropriate airports with high volume and complex traffic flow. On a broader scale, the Safety Board encourages the FAA to continue the research effort in Airport Surface Traffic Automation (ASTA), which is intended to develop automation tools and more complete automation for controlling the flow of aircraft on the airport surface. In addition to reducing the frequency of runway incursions, design goals of the program should include a reduction ih taxiway incursions and improvements in ATC operational efficiency.

[ANALYSIS] However, during an additional conspicuity exercise, it was visually evident from both the tower and the final approach that the aircraft and runway lights tend to blend together, perceptually. During the field phase of the investigation, members of the Safety Board’s technical staff, with support from representatives of the airline industry and the FAA, conducted an aircraft external lighting detection task/exercise at LAX during night visual meteorological conditions (VMC). A Metroliner identical to the one involved in the accident was placed at the same location on runway 24 left where the collision occurred. The airplane was aligned with the centerline of the runway and its navigation and anticollision lighting were on and operating. The runway edge lighting and centerline lighting were at low (step 2) intensity. During visual approaches to the runway, cockpit observers found it difficult to differentiate between the Metroliner and the lighted runway environment. The size of an aircraft and its proximity to the runway lighting, espectally on runways with centerline lighting, make these light sources virtually indistinguishable when viewed from directly behind and above. The visual approach exercises also indicated that the likelihood of detecting an aircraft from the rear on an active runway by an approaching 60 aircraft can be increased if the first aircraft is displaced from the runway centerline lighting by approximately 3 feet. Moreover, when this offset procedure was used in conjunction with high-energy strobe lighting and anticollision and navigation lighting, aircraft conspicuity was enhanced. The Safety Board notes that most air carriers, and a considerable number of general aviation aircraft operating in the National Airspace System (NAS), are equipped with some form of high-energy strobe lighting. Therefore, this combination of actions, as well as equipment, would be available to nearly all users in the NAS. ° Officials from the Aviation Safety Reporting System of the National Aeronautics and Space Administration (NASA) have conducted’ several analytical studies of reports by pilots and controllers involved in runway transgressions. The latest study, published in 1985, revealed that the most frequently cited factor in controller-enabled departure transgressions was "controller failure to visually locate traffic." The Safety Board believes that the use of strobe lighting, along with the practice of displacing the aircraft off the centerline lighting, would significantly enhance the ability of pilots and air traffic controllers to visually detect traffic conflict situations. The use of strobe lighting by aircraft occupying an active runway would also ease the controllers’ memory load by assisting them in locating, identifying, and segregating aircraft on an active runway. During the Safety Board’s public hearing on the Los Angeles accident, testimony was received from representatives of the FAA and industry concerning aircraft external lighting standards and conspicuity.

[ANALYSIS] From experience, the controllers expected commuter airplanes departing from the north runway complex to request midfield departures either from runway 24 left or 24 right. Weather conditions were well above the criteria for VFR. In postaccident interviews, neither the surviving flight crewmember of USA1493 nor the air traffic controllers identified environmental factors as a constraint to the normal performance of their duties. The physical evidence on the surface of runway 24 left at the intersection of taxiway 45 and the witness marks on the surfaces and structure of both airplanes indicated that the collision occurred on a runway that was the responsibility of the LC2. 2.2 Air Traffic After the crew of SKW5569 had received the flight plan clearance from the controller at Clearance Delivery in accordance with local procedure, the flight strip for the flight was forwarded directly to the LC2 position. Because the boarding gates for Skywest Airlines are on the south side of the airport at terminal 6, the flightcrew received initial taxi instructions from the GCl (south complex) ground controller. Due to the northeastbound route of flight, the airplane was cleared to proceed to the north route via taxiway 48 and made initial contact with the GC2 (north complex) ground controller at the appropriate changeover point. The flightcrew was then instructed to taxi to runway 24 left. 50 In an effort to reduce workload at the ground control position, LAX ATC procedures did not specify the use and handling of flight progress strips at that position. As a result, aircraft could request intersection departures directly from the local controller. The ground controller was thereby relieved from coordinating with the local controller and marking flight progress strips accordingly. Although intended to reduce the ground controller’s workload, the procedures eliminated redundancies that were built into the system and increased the local controller’s workload. Without the flight progress strip information, the local controller was required to determine the flightcrew intentions and rely on memory and observations of aircraft moving on the ground to identify and track the progress of aircraft under his/her control. If a controller is unable to recall such details or unable to observe or recognize an aircraft, however briefly, the possibility of error is greatly increased. A review of the communications transcript of the LC2 position provided the following insight regarding a previous airplane’s request for an intersection takeoff: When SKW246 advised, "two forty six will take forty seven," the response, "hold there," indicated that she was aware of this particular aircraft’s position. This awareness is again apparent when she asked the flightcrew of SKW246, "...you still holding short of forty seven?" When she received an affirmative response, she advised the flightcrew, "you’re next," indicating her intention to take specific action with this flight after the departure of USA23, which she had just cleared for takeoff on runway 24 left.

[FINDINGS] Air traffic volume and traffic control workload at the Los Angeles International Airport was moderate at the time of the accident. Weather conditions did not contribute to the cause of the accident. The ability of the Los Angeles Air Traffic Control tower personnel to distinguish aircraft on the runways and other airport traffic movement areas, including the accident site, was complicated by some of the terminal II apron lights which produced glare. Operating procedures at the Los Angeles Air Traffic Control tower did not provide redundancy comparable to the FAA’s National Operational Position Standards, which require that flight progress strips, used to monitor the progress of flights between controller positions, be processed through the ground control position. FAA evaluations, as administered by the Air Traffic Service staff, did not identify that essential redundancy was absent at the Los Angeles Air Traffic Control tower. This lack of redundancy contributed to and compounded errors by the local controller. The local controller forgot that she had placed SKW5569 into position for takeoff on runway 24 left at the intersection of taxiway 45 because of her preoccupation with another airplane. The local controller’s incorrect perception of the traffic situation went undetected because she had an apparent match between her view of the traffic situation on the airport and the flight progress strip at her operating position. ll. 12. 13. 14, 15. 16. 17. 18. 19. 75 A flight progress strip for WW5072 was earlier misplaced by the clearance delivery controller. If local procedures had required that strips be processed through the ground control position, misplacement would have been detected and corrected. Because this strip was not present at the local controller’s operating position, she misidentified an airplane and issued a landing clearance that led to the runway collision. Current communications procedures for pilots and controllers regarding intersection takeoffs do not require that a specific point of departure be identified. The Technical Appraisal Program for air traffic controllers is not being fully utilized because of a lack of understanding by supervisors and the unavailability of appraisal histories. The local controller did not have the Airport Surface Detection Equipment radar available to assist her; however, under the circumstances and procedures in effect, it probably would not have prevented the accident. Aircraft external lighting systems required for certification are intended primarily for in-flight conspicuity, rather than for conspicuity on airport surfaces; consequently, the external lighting of SKW5569 tended to be indistinguishable from the runway lights when viewed from the cockpit of USA1493. The postmortem presence of phenobarbital in the captain of USA1493 and over-the-counter medications in the first officer of SKW5569 did not contribute to the accident. However, it indicates a less than complete appreciation of the potential dangers that the unauthorized use of such medications may pose. The emergency response of the Los Angeles Department of Airports for this accident was timely and effective. The exit row briefing provided by USAir increased the preparedness of passengers for the evacuation; however, the delay in opening the right overwing exit, the partially blocked exit opening and other reaction to stress caused delays in the egress of some passengers.

[ANALYSIS] The use of strobe lighting by aircraft occupying an active runway would also ease the controllers’ memory load by assisting them in locating, identifying, and segregating aircraft on an active runway. During the Safety Board’s public hearing on the Los Angeles accident, testimony was received from representatives of the FAA and industry concerning aircraft external lighting standards and conspicuity. An FAA lighting specialist testified that the federal standards for aircraft external lighting are primarily intended to serve in-flight conspicuity needs and that no effort has been made by the FAA to address the issue of conspicuity of aircraft on airport surfaces. The Safety Board believes that the FAA should study and evaluate ways of enhancing the conspicuity of aircraft on airport surfaces during night or periods of reduced visibility. The concept of displacing an aircraft away from the centerline lighting and the use of lighting enhancements, such as high-energy strobe lighting and logo lighting, by aircraft on active runways should be explored and evaluated for their value to the conspicuity issue. A representative of the Fairchild Aircraft Company, the manufacturer of the Metroliner, testified that the flightcrew of USA1493, due to line-of-sight obstruction, may have been unable to see the anticollision beacon on top of the vertical stabilizer. The Metroliner’s rudder cap obstructs the beacon when viewed from the rear. As the flight descended below 100 feet over the runway surface, "it is very possible he couldn’t see the beacon." When the surviving flight crewmember of USA1493 was asked to account for the fact that he didn’t see the Metroliner earlier, he testified, "It wasn’t there. It was invisible." 6] Federal Aviation Regulations permit some aircraft structural obstructions, which, in this case, interfered with the flightcrew’s ability to see the anticollision beacon. Nevertheless, the anticollision beacon obstruction on N683AV was within the allowable criteria. The Safety Board has been unable to determine with certainty whether the inability of the flightcrew to detect the anticollision beacon when USA1493 was below 100 feet over the runway surface contributed to the accident. Nevertheless, the Safety Board believes that in establishing permissible areas of obstruction, the coverage compliance standards should give consideration to the approach, overtaking, and takeoff situations; that is, the anticollision light of an aircraft in position on a runway should be clearly visible to the pilot of another aircraft planning to land or take off on that runway. The Safety Board therefore believes that the FAA should reevaluate and redefine the permissible areas in which the illumination of an anticollision light is obstructed by aircraft structure. The intensity and vertical coverage of the anticollision beacon on N683AV met the performance standards under which the airplane was certificated. The Safety Board is aware that airplanes certificated after September 1, 1977, are required to have an anticollision light with an intensity of 400 candles and a vertical coverage of 75 degrees above and below the horizontal plane of the airplane.

[ANALYSIS] The study did not evaluate controller training or human performance issues. The study did suggest that incident reporting might be part of the problem since there were indications that not all incidents are reported, which caused a situation that precluded appropriate corrective measures. Furthermore, the report did not propose any specific corrective measures. Although the FAA did conduct the study on the runway incursion problem, the study did not result in the development of remedia! action to reduce or alleviate the problem. The Safety Board’s concern about the problem was heightened again after it investigated a near collision between two DC-10’s at the Minneapolis-St. Paul International Airport on March 31, 1985. That occurrence prompted the Board to undertake a special investigation study of runway incursion incidents. Since that time, additional incidents and Sworth Central Airtines, Inc., Medonneli Dougtas OC-9-31, NO54K and Detta Air Lines, Inc. Convair CV-880-NS807E, O'Hare [International Airport, Chicego, Litinois, December 20, 1972. MISB/AAR-73-15. Ssince 1972, the Safety Board has issued 42 s<:fety Recommendations to the FAA addressing the probles of runway incursions. A summary of these recoamendations and theic current status fs contained in appendix F. TWideir Cot'ision Invotving a Falcon Jet, %1291GW, and Cessna 150, W6423K, Memphis, Tennessee, May 18, 1978. MNISB/AAR-78-184. 38 accidents have continued to occur,® which suggests the need for other measures to resolve this problem. During 1985 and early 1986, the Safety Board investigated 25 more runway incursion incidents that were summarized in a special investigation report adopted on May 6, 1986.% The Safety Board found that the incursions were the result of both controller errors and pilot deviations. The report indicated that controller operational errors generally resulted from a coordination breakdown between local and ground controllers or distractions that diverted a controller’s attention from a developing or established conflict situation. Pilot deviations accounted for about 30 percent of the incursions investigated by Safety Board staff and involved misinterpretations of clearances and unauthorized runway crossings. Many of the incursions could have been avoided, if the pilots had looked for traffic before proceeding onto an active rurway. Fourteen recommendations were sent to the FAA addressing issues, suth as procedures, training, pilot and controller communications and airpert signing. The FAA took several actions in response to the Board’s recommendations. These actions included establishing a runway incursion data base, distributing training material, including a video tape to bring controller and pilot attention to the problem, changes to controller and pilot phraseology, and placing more emphasis on airport taxiway guidance Signs.

[ANALYSIS] Fourteen recommendations were sent to the FAA addressing issues, suth as procedures, training, pilot and controller communications and airpert signing. The FAA took several actions in response to the Board’s recommendations. These actions included establishing a runway incursion data base, distributing training material, including a video tape to bring controller and pilot attention to the problem, changes to controller and pilot phraseology, and placing more emphasis on airport taxiway guidance Signs. The MITRE Corporation also conducted an analysis of runway incursions summarized in a report in April 1989. This report'® defined controller-related factors as follows: 0 Erroneous scanning, or failure to scan the runway or approach path (local controller [LC) and ground controller (GC)). Forgetfulness about the traffic situation (LC and GC). Misju‘jment of traffic separation (LC). Scottision in vreroit, Michigan, December 3, 1990, between Northwest flights 299 and 1482, and Collision in Los Angeles, Catifornia, February i, 1991, between USAi« 1493 and SkyWest 5567 are under NITS8 investigation. 9addirtionat information on the generat subject of runway incursions can be found in the National fFransportation Safety Board, Special Investigation Reports, “Runway [Incursions at Controtled Airports in the United States," WISB/SER-86/01, May 1986, and am FAA pubticattion "Reducing Runway Incursions: An FAA Report" dated April 1990. Othe MITRE Corporation. “An Analysis of ATC-Related Runway incursions, with Some Potential Technoltogicai Solutions,” MIR-89W9I021, Aprit 1989. 39 Lack of coordination or inadequate coordination with the LC on runway crossings (GC}}. Errors in sending or receiving clearances and instructions (LC and GC). The runway collision of EA 111 and King Air N44UE involved these same controller-related factors identified by both Safety Board and the MITRE corporation studies. These factors are human performance-related and are being addressed in a number of different actions, including FAA and industry efforts to increase awareness of the magnitude and nature of the human performance problem, improved training and technological solutions that may reduce the workload, and a fail-safe redundancy for the human performance of air traffic controllers. The Safety Board is aware of several advanced concepts in airport surface traffic detection and automation that, when perfected with the correct match of hardware and location specific software, could provide warnings to preclude accidents of a nature similar to the collision of €A 11] and N44UE. For example, the FAA is currently testing an Airport Movement Araa Safety System (AMASS). The AMASS system will use the data available in ASDE-3 and the ARTS to identify potential incursions and will alert the controller so that timely corrective actions can be taken. The Safety Board fully supports the early development and installation of such systems at airports where the volume and complexity of traffic flow dictates its use.

[ANALYSIS] Therefore, the Safety Board believes that the procedures contained in the Air Traffic Control Handbook, 7110.65F, paragraph 3-127, "Anticipating Separation," should be amended to preclude the issuance of multiple janding clearances to aircraft outside of the final approach fix. Also, a numerical limit should be established so that n° more than two landing clearances may be issued to successive arrivals. The Safety Board believes that this change will increase system effectiveness, while not creating an undue burden on the controller. Nevertheless, pilots also have a responsibility for separation assurance on the runway and vigilance during landing must be a shared. The Safety Board is aware that if the local controller had provided traffic information to the crew of EA 111, the accident might have been prevented. This procedure, had it been followed, would probably have prompted the crew to query the local controller as to the position of their traffic on the runway, since it was unlikely that visual observation would have occurred. As a_ system redundancy, the Safety Board believes that the importance of issuing traffic information to arriving aircraft should be stressed. Therefore, the Safety Board believes that a mandatory, formal briefing should be provided to all air traffic controllers on the importance of, and the need for, giving traffic information when issuing an anticipated separation landing clearance. The briefing should be contained in an Air Traffic Bulletin. 2.5 Efforts to Reduce Runway Incursions The Safety Board’s concern about the hazard of runway incursions dates back to 1972 following an accident at the Chicago O’Hare International 37 Airport.> As a result of that accident, four Safety Recommendations were issued to the FAA addressing air traffic control procedures and pilotcontroller communications.® The Board’s concerns were further reiterated in 1979 following two more runway incursions incidents and one accident.’ These occurrences prompted the Board to recommend that the FAA conduct a directed safety study to examine the runway incursion problem and to formulate recommended remedial action to reduce the likelihood of such hazardous conflicts. That recommendation was issued in June 1979. In response, the FAA commissioned the Transportation Systems Center in Cambridge, Massachusetts, to conduct a stucy. The study was completed in Apri) 1981 with a report. entitled "An Analysis of Runway-Taxiway Transgressions at Controlled Airports." The study concluded that “there does not appear to be any pattern to the causes . . . other than human errors on the part of both air traffic controllers and pilots." The study also concluded that “more uniform communication and verification of messages between pilots and controllers could serve to reduce the chance of ambiguous or erroneous commands/actions." The report raised the question as to whether system reliability might be improved by increasing the reliability of the human element or by adding redundant elements. The study did not evaluate controller training or human performance issues. The study did suggest that incident reporting might be part of the problem since there were indications that not all incidents are reported, which caused a situation that precluded appropriate corrective measures.

[ANALYSIS] The CO 9687 fligh:crew took positive action to avoid becoming a hazard. They stopped clear of the active runways and remained in position until their clearance was clarified. 34 The Safety Board conciuded that there was a lack of understanding between the tocal controller and CO 9687, and that this communications anomaly did not result from any oquipment failure. Rather it was the result of an incomplete transfer of intormation (taxi instructions) between the controller and the flightcrew of CO 9867. The net effect of this lack of information transfer was to create a self-imposed workload on the controller that was sufficiently high to cause him to disregard other higher priority tasks. If the local controller had been so concerned that the airplane was going to cross runway 26 left without a clearance, he had the option of discontinuing departures from that runway and directing ASE 301, that was holding in the takeoff position, to clear runway 26 left. As long as ASE 301 was holding in position on the runway, it could not be threatened by a possible runway incursion from an airplane at the opposite end of the runway. However, the local controller became distracted for a critical period by the possibility of a runway incursion involving ASE 301 and CO 9687. Eventually, et 1904:13, the local controller cleared ASE 301 for takeoff on runway 26 left. The collision of €A !1] and N44UE had taken piace on runway 26 right at 1904:07. It is recoynized that concentration on one task can overload a person to the extent that other relevant cues are disregarded or otherwise not attended to, leading to a degradation of overall task performance. Because the north local controller focused his attention on the path of CO 9687 on taxiway Bravo, at the west end of runway 26 left, he was distracted at a critica) time from the landing rollout of N44UE and the FA 111 airplane that was about to cross the threshold and land on the same runway. The Safety Board concludes that this accident was a result of lapses in the performance of the Atlanta tower north tocal controller and, to a lesser extent, the performance of the Atlanta approach control north final controller and the radar monitor controller. Specifically, the north local controller did not ensure the separation of the aircraft approaching and landing on runway 26 right. further, ne failed to follow the prescribed procedure of issuing appropriate traffic information to the crew of EA 11). This information would have improved the flightcrew’s situational awareness and their motivation to search for the preceding King Air. The Atlanta approach north final controller and the radar monitor controller had opportunities to issue timely speed reductions to ensure that adequate separation was maintained between the successive aircraft on final approach, but did not do so. Although these lapses of controller performance are cited as causal, the Safety Board also has chosen to recognize that the controllers’ performance was a direct product of FAA air traffic management institutional decisions and practices that do not allow for human performance lapses in judgement or decision making.

[ANALYSIS] No indication was given that they were number two for landing behind another airplane. The Safety Board believes that flightcrews are conditioned to receive such information. as required in the Air Traffic Con*rol Handbook procedures relating to anticipating separation. If the controller had provided traffic information to the EA 11] flightcrew, the flightcrew’s sense of situational awareness and motivation to search for a preceding airplane might have been increased. Lacking such information, it appears that the crew proceeded through their normal task of completing a routine night landing on a runway to which they had been cleared, unaware that there was another airplane on the runway. The fact that EA 111 had received a landing clearance did not relieve the flightcrew of responsibility to "see and avoid" other aircraft in their vicinity. However. in the absence of conspicuous lighting on the King Air and without prompting from ATC to direct their attention to traffic ahead, it was extremely difficult, if not impossible for the EA 11] flightcrew to detect the other aircraft on the runway. Moreover, there is a concept known as diffusion of responsibility that describes a tendency on the part of pilots in some circumstances to relax their vigilance. A National Aeronautics and Space Administration study on near midair collisions‘ sBillings, C., Greyson, ®., Hecht, W., and Curry, R., "A Study of Near Midair Collisions in U.S. Terminal Airspace," NASA Technicat Memorandum 81225, 1980. 32 indicates that an inappropriate sense of shared responsibility may occur when an airplane is under ATC radar control. In such a circumstance, a pilot may relegate a portion of his responsibility for vigilance to the controller for seeing and avoiding other aircraft. In the case of EA 111, having come from the radar environment of the approach and after having received specific landing clearance, the pilots may have experienced a natural tendency to relax in their attempts to visually search for an aircraft between their position and the intended landing runway. In any event, the Safety Board found no evidence to indicate less than expected vigilance by the EA 1}}] flightcrew. The Safety Board thus concludes that the actions of the EA 111 flightcrew, while not optimal in terms of speed control and situational awareness during initial and final approaches, were not uncommon to airline operations and were not causal to the accident. 2.4 Role of Air Traffic Control The Safety Board also evaluated the performance of the air traffic control personnel involved in this accident. The final controller was responsible for maintaining separation of succeeding airplanes on the approach to the outer marker. The monitor controller was responsible for maintaining separation of succeeding airplanes on the approach from the outer marker to within 1 mile of the runway.

[ANALYSIS] Provision of the multiple clearances to land in retatively rapid succession may have provided the north local controller with time needed to devote attention to flights waiting for takeoff clearances from runway 26 left. However, the premature clearances also make the north local controller the only controller responsible for the spacing between successive flights by removing that responsibility from the monitor controller and it may also have reduced the vigilance of the flightcrew. As a result, appropriate spacing for completion of the landings depended entirely on the continued vigilance of the local controller and the flightcrew. In this case, the flightcrew of EA 111 probably could not have seen N44UE because of N44UE’s external lighting configuration. Also, the premature clearance for EA 111 to land removed the redundancy of flightcrew vigilance when the north local controller subsequently became distracted with 36 CO 9687’s taxi clearance. Without the clearance to land, EA 111 would have had to remind the local controller that the clearance was needed. This reminder would probably have redirected the controller’s attention to tte lack of adequate spacing between N44UE and EA 11] and may have led to a correction of the problem by denying EA 11] clearance to Jand. In the 1977 edition of the Air Traffic Control Handbook, 7110.65/, the issuance of multiple landing clearances was not allowed. Specifically, paragraph 1122, “Anticipating Separation,“ stated, “Landing clearance need not be withheld until prescribed separation exists if there is reasonable assurance it will exist when the aircraft crosses the landing threshold. However, do not clear a succeeding aircraft to land on the same runway before a preceding arriving aircraft crosses the landing threshold...." This is basically the same text that is contained in the current Handbook, 7110.65F; however, the earlier procedures went on to say, "...do not clear more than the first two aircraft to land at any one time ai include traffic information with the clearance." During March 1978, isis paragraph was changed to delete the numerical limits for clearing aircraft to land. The Safety Board believes that current ATC procedures, as they oertair to the anticipated separation of arriving aircraft, require nearly flawless human performance that makes no allowance for an error of omission or lepse of attention due to any type of distractive event. Therefore, the Safety Board believes that the procedures contained in the Air Traffic Control Handbook, 7110.65F, paragraph 3-127, "Anticipating Separation," should be amended to preclude the issuance of multiple janding clearances to aircraft outside of the final approach fix.

[ANALYSIS] At the time of the north local controller’s transmission, EA 111 was almost 6 miles from the runway and the King Air was about 3 miles out. However, the distance between the two aircraft was decreasing at an unacceptable rate and was less than the required 2.5 miles separation as N44UE arrived at the runway threshold. The north local controller, in an attempt to maintain the lancing sequence, initiated visual separation between N44UE and EA 11]. At the time visual separation was initiated, the required minimum radar separation standard of 2 1/2 miles did exist between N44UE and €A 11:. However, to make Sure that an approved separation standard would exist after using visual separation, the local controller would have had to monitor both airplanes closely to assure that EA 11] did not cross the runway threshold until N44ue had been observed leaving the runway (Air Traffic Control Handbock 7110.65, paragraph 3-122, Same Runway Separation}}. Unfortunately, EA 111 had about a 45 knot closure rate on N44UE, and ATC radio transcripts indicate that no action was taken to reduce the rate of closure. In addition, the radio transcripts indicate, and a personal interview confirmed, that the local controller became distracted by radio communication difficulties with the flightcrew of CO 9687. The local controller stated to investigators that he did not observe the touchdown and rollout of N44UE at the runway threshold or during landing. The Safety Board reviewed the pertiient FAA Air Traffic Control Handbook 7110.65F requirements and concluded that the instructions contained therein clearly define the controller’s responsibilities for “same runway separation" and “anticipating separation.” The Safety Board concluded that the physical evidence on the runway and on both airplanes indicated that the collision occurred on a runway that was the responsibility of the narth local controller. The Safety Board attributes the north local controller’s distraction and preoccupation with efforts to communicate with CO %687 to his gerceived need to clear runway 26 left for another airplane inbound with a hydraulic emergency. However, the Board believes that the local controller’s concern that the flightcrew of CO 9687 was going to cross runway 26 ieft without a clearance was not well founded. At 1903:03 radio transmissions indicate communications difficulties between CG 9687 and the tower regarding its taxi instructions. They transmitted, "tower, Continental ninety six eighty seven bravo two [taxiway] holding short." The fact that the airplan7 was on the taxiway and not moving was substantiated by the ground controller in the tower cab. The CO 9687 fligh:crew took positive action to avoid becoming a hazard. They stopped clear of the active runways and remained in position until their clearance was clarified. 34 The Safety Board conciuded that there was a lack of understanding between the tocal controller and CO 9687, and that this communications anomaly did not result from any oquipment failure.

[FINDINGS] Visibility from DL 349 toward the landing area of runway 9R was 3,000 ft. DL 349 entered runway 9R at » magnetic heading of 128° from a high-speed turnoff approximately 4,000 ft from the runway threshold. From a normal reference eye position, the captain of DL 349 could not have seen the FT 74 until I to 2 see before the B-747 passed in front of his aircraft. From the ncrinal eye reference position, the first officer of DI. 349 could have seen FT 74 about 4 sec before the near-collision. Passengers of DL 349 saw FT 74 while it was in the air and on the runway coming toward them. Because the corimunication radios of each aircraft were tuned to different frequencies, the ground controller's transmissions to DL 349 were not heard by the FT 74 crew and the FT 74's transmission for DL 349 to stop crossing runway 9R was not heard by the DL 349 crew. 3.2 Probable Cause The National Transoortetion Safety Board cetermines that the probable cause of this accident was the O'Hare outbound ground controller's issuance of a taxi clearance: across runway 9R, which permitted Delta Flight 349 to move into a collision path with Flying Tiger Flight 74 anc, further, the failure of the pilots of Delta Flight 349 to maintain a continuous vigil for landing traffic before entering an active runway. The improper clearance was the result cf the ground controller's failure to see the displayed radar taryet of the landing aircraft. Contributing to the accident were the appreach controller's failu:e to effect required spacing criteria between Flying Tiger Flight 74 and the preceding arrival aircraft and the local controller's failure to issue a missed approach clearance when he noted the less-than-required separat! :n. As a result of this accident and other runway incursion incident/accidents which oceurred at La Guardia Airport, New York, and Memphis Airport, Memphis, Tennessee, the National Transportation Safety Boarc recommended on June 8, 1979, that tne Federal Aviation Administration(See Appendix H): "Conduct a direeved safety study, on a priority basis, to examine the runway incursion problem and to fortnulate recommended remedial action to reduce the likelihood of such hazardous conflicts. (Class Il, Priority Action) (A-79-42) "Alert all controlles/pilot personnel that runway incursion mishaps represent & serious safety problem which requires their immediate attention. Speciel emphasis should be placed on the need for both groups to maintain greater visual surveillance in those taxi operations involving any runway crossing. (Class U, Priority Action) (A-79-43)" BY THE NATIONAL TRANSPORTATION SAFETY BOARD /s/ JAMES B. KING Chairman /s/ PATHICIA A. GOLDMAN Member /s/ GH. PATRICK BURSLEY Member ELWOOD T. DRIVER, Vice Chairman, did not participate. FRANCIS H. McADAMS, Member, filed the following dissenting statement.

[PROBABLE CAUSE] The Board has completely missed the point of this accident, since even if Delta had visually checked the runway at 6910:18 when Flying Tiger was 900 feet down the runway, Delta could not have seen Flying Tiger because of the restricted visibility and Delta would have entered on the runway as previously cleared. In other words, ATC vectored two aircraft on a collision course on the runway, and the attempt to blame Delta for being on the runway is highly unreasonable under the circumstances, The Board s-ates: From the normal eye reference position, the first officer could not have seen FT 74 until about 4 seconds before the near collision. Although the response time was minimal, it probably was sufficient to have Stopped the aircraft and permitted FY 74 to pass safely onto the runway. At this time Delta was almost to the midpoint of the runway, and a potentially dangerous situation now existed. Even if Delta had stopped, an accident or nearcollision would have already occurred; Flying Tiger weuld have had to swerve to the right to avoid a collision in any event. Under these circumstances I find, contrary to the majority, that stopping the aircraft would not have avoided an accident or incident. A pilet receiving positive clearance to cross an active runway should visually clear the ranway for landing traffic if he ean physically see it. On the other here, in this case the ground controller should have been aware of the restricted meteorologicel conditions and not have issued the clearance. According to the majority's reasoning, Delta should not have crossed the runway until it was possible to visually clear the runway and appreach. Unfortunately, it was not possible to visually clear the runwav until there was a substantiel improvement in the visibility conditions, Under these circumstances, Delta had the right to rely upon end accept the radav-vectcred instrument taxi clearance to cross the runway, just as Flying Tiger had the right to rely upon its landing clearance. However, of tar more significance than the foregoing is tne fact that once Delta had turned to a heading of 118 degrees to cross runway 9R it would have been physically imnossible for Delta to have seen Flying Tiger, even if there hed been no restrictions to visibility. Flying Tiger was not within the visual enveiope of Delta until 0910:25; at. 9910:25 the Delta aircraft had intruded onto runway 9. At all times price to 0910:25, Flying Tiger was behind the right shoulder of the Delta first officer ut about the 4:30 o'clock position. The facts appear to be clear. At 0910:19, Delta was on a magnetic heading of 118 degr2es, and the nose of the aircraft was near the north edge of runway 9. At this time Flying Tiger was 55 feet in the sie and about 2,800 feet from Delta. It would have been physically impossible for the Delta first officer to have seen Flying Tiger until 0910:25 — even if he had been leaning forward in his seat.

[ANALYSIS] It was not until the 8-747 was within 150 tt that the DL 349 captain could have seen the other sireraft. The Sefety Board notes that 14 CFR 913 holds the pilot in command of an aircraft to be directly responsible for, and the final authority as to, the operation of thst aircraft. Further, the Airman's Information Manual (AIM) 8/ states that ATC clearances and instructions pertaining to taxiing are predicated on known traffic and known airport conditions, and that although an ATC clearance is issued for taxiing purposes when operating in accordance with the FAR's, it is the responsibility of the pilot to avoid collision with other aircraft. In this accident, the captain of DL 349 failed to clear his right of way, and because he could not adequately scan for landing aircraft, he should have told his first officer to look for traific before entering on the active runway. If he had done so, the accident would have been prevented. 3. CONCLUSIONS Findings 1, The ORD outbound ground controller did not effect coordination with the local controller before clearing DL 349 acress the active landing runway, nor was he required to do so. An ORD tower directive, ORD Tower Order 7110.7B, authorized the ground controler to clear aircraft or vehicles across an active runway based on his independent judgment and without local controller coordination. To comply with the order, the ground controller must use the ASDE and the BRITE displays and observe the landing runway before issuing a taxi clearance to cross an active runway. The approach controller did not provide the required separation between UA 225 and FT 74, The monitor controller was unable to increase the spacing. The outbound ground controller probably anticipated tnat the next arrival, sequenced behind UA 225, would be about 4 mi in trail. Based upon what he perceived to be a safe interval between landing traffic, the ground controller cleared DL 349 across runway 9R and into the path of FT 74. The ground controller had failed to see the displayed radar target of FT 74 on the BRITE display. At the time of the controller's crossing clearance, FT 74 was not in visual range of the controller nor was it within the ASDE surveillance range. 87 AIM, Paragraph 241(b), January 1979. The DL 249 flighterew looked for landing aircraft on runway 9R as their aircraft turned on the 14R parallel taxiway. They did not maintain a traffic lookout as they continued taxiing. Visibility from DL 349 toward the landing area of runway 9R was 3,000 ft. DL 349 entered runway 9R at » magnetic heading of 128° from a high-speed turnoff approximately 4,000 ft from the runway threshold. From a normal reference eye position, the captain of DL 349 could not have seen the FT 74 until I to 2 see before the B-747 passed in front of his aircraft.

[ANALYSIS] If the approach and local controllers, backed bi ‘ale monitor controller, had sequenced FT 74 at a 4-mi landing intervai, the grourd controller would have had adequate spacing to clear DL 349 eecross the activ. landing runway after UA 225 passed the runway 14R and runway 9R intersection. Chicago-O'Hare Tower Order 710.7B authorized the ground controller to clear traffic across an active runway without ccordinating with the ‘ocal controlier. The effectiveness of the directive is contingent on the proficiency of the ground controller to observe the activities of the local controller, to perceive the traffic flow, scan the runways, and monitor the ASDE and the BRITE display. The ORD order authorizes him to make an independent judgment as to whetter it is Safe and expedient to clear an aircraft or vehicle across an active runway. The ground controller had demonstrated his ability to move ground traffie sefely in accordance with the ORD Tower Order many times during his 2 years at OF.D. The outbound ground controller did not effect coordination with the locel controller before issuing the clearance to DL 349 tc cross the active landing runway. Although he was not required to Jo so, the local controller should have advised the ground controller of the considerably reduced horizontal spacing between UA 225 and FT 74. If he had been so advised, the ground controller would have had sufficient information to preclude an improper judgment regarding the actual spacing between the aircraft arriving on the ILS approach course. This accident illustrates that there was a deficiency in the local order in that it did not provide for the probability of human error. On March 22, 1979, Chicago-O'Hare Tower Order 7110.7C was issued which preseribed improved coordination between ground and local controllers when aircraft cross active runways. The ground controller stated that he checked the BRITE display for landing traffic before he cleared DL 349 to cross runway 9R. He observed a radai: target at "an estimated distance of three and one-half mi" behind UA 225 but he did not read the associated alphanumeric data tag. In as much as there was no target at a distance of 3.5 mi, the radar target that he saw was TW 291, which actually was about 5.25 mi from the runway threshold. The ARTS II coraputer at ORD is prorrammed to display data tags for transponder-equipped arrivals to a point 2 mi from the approach end of runway 9R. At this point the data tag is automatically dropped and the basic radar target is cinphasized with an asterisk overlay as the taryet continues toward the runway threshold.

[PROBABLE CAUSE] KING Chairman /s/ PATHICIA A. GOLDMAN Member /s/ GH. PATRICK BURSLEY Member ELWOOD T. DRIVER, Vice Chairman, did not participate. FRANCIS H. McADAMS, Member, filed the following dissenting statement. I diseg'ree with the majority of the Board wherein they conclude inter al’. that the probable cause of the accident was". . .the failure of the pilots of Delta Flight 349 *o maintain a continuous vigil for landing traffic before entering an active runway." The facts are as follows: Delta was cleared to cross the runway at 0910:00 by the outbound ground controller and told to "keep it moving." At this time Delta was approximately 472 feet from the runway. Shortly after receiving clearance to cross, both Detta crewmembers looked toward the approach end of the runway and observed no treffic. The approach to the runway, as well as approximately 950 reet of the approach end of runway $R, was obscured by a fog bank; consequently, *{{ was not possible for the Delta crew to observe any approaching traffie until it was at least 950 feet from tne runway threshold. The Delta aireraft continued to taxi onto the runway, and when the nose had intruded to about 75 feet onto the runway the Flying Tiger aircraft was observed by the Delta first officer at about 0930:29. The near-miss occurred at 6910:31. Even according to the majority, the first time that the Deita first officer could have seen Flying Tiger from his normal coekpit position was at 0910:27, or 4 seconds before the near-collision, Delta was almost to the midpoint of the runway at this time. The local controller stated that due to the existing visibility conditions he first observed landing aircraft as they touched down at the glide slope intersection point, 1,200 feet from the threshold. Flying Tiger passed this point at 0910:20, and ~-20- the nose of the Delta aircraft had already intruded onto the runway. In this conncetion, it is significant that the Flying Tiger crew did not see Delta until C910:27, or 4 seconds before the accident—about the same time that Delta observed Flying Tiger. Ba7ed on these facts, a majority of the Board has concluded that, despite an ATC clearance to cross the active runway in severely limited visibility conditions, the Delta crew could have avoided the accident if a continuous vigil for ianding traffie had been maintained. The Board has completely missed the point of this accident, since even if Delta had visually checked the runway at 6910:18 when Flying Tiger was 900 feet down the runway, Delta could not have seen Flying Tiger because of the restricted visibility and Delta would have entered on the runway as previously cleared.

[ANALYSIS] Several minutes befcre the accident, the local controller had advised landing aircraft that runway visual range for runway 9R was 3,900 ft in the touchdown zone and 6,000 ft in the roliout area. He also observ>d the RVR on cunway 9R as 3,000 ft when UA 225 landed although he did not report, as required, the RVR reading to FT 74. He stated that he first saw the majority of landing aircraft as they touched down at the glide slope intersection point. The runway 9R transmissometer was located closer to the approach end of the runway than the glide slope intersection point and it was at this point that the 3,000-1t RVR vaiue was taken. The Safety Board recognizes that the visibility along a RVR baseline is not always representative cf the visibility outside the sampling volume 7/, and we believe such variable visibility conditions existed on the airport at the time of the accident, It is note\.orthy that the DL 349 captain recalied that a fog bank was off the end of runway 9R. The captain of FT 74 confirmed the fog's presence when he described a "white-out" area as he descended below the overcast. As outside visual reference was limited, he continued the descent to the runway by cockpit ILS instrument reference. Regardless of the different visibility distances observed i7 Analysis of Visibility Ob.ervation Methods, Hockreiter, 1969, U.S. Department of Commerce. ~16- at varicus airport locations, the Safety Board concludes that the runway 09R RVR accurately reflected the 3,000 ft visual range which was “«‘lable to the DL 349 pilots in the direction of the landing B-747. The cockgit visibility study determined that the visuel angle in the horizonta: plane, with the pilots’ eyes in the normal reference position, permitted the captain to see objects within a 115° are to the right from a point directly in front of him and per aitted the first offieer to see within a similar 137° are to the right, The visual envelope would have been increased if either pilet had leannd forward or to his right. After acknowledging the controller's crossing clearances, both the Delta captain and first officer looked toward the apprcach end of runway 9R as their aireraf: completed the turn onto 14R parallel taxiway, a point which is 250 ft from the north edge of runway 9R. The first officer had 6n unobstructed view toward the runway threshold; however, his visual range ended 9° a point about 95u ft from the threshold. The captain could have seen the same distance only if he had ‘eaned froni his normal eye position, which he said that he cid. The B~-747 was beycnd their visual range. At 0910:19, FT 74, while 55 ft above the runway, was about 2,800 ft from DL 349 and within visual range.

[PROBABLE CAUSE] At 0910:19, Delta was on a magnetic heading of 118 degr2es, and the nose of the aircraft was near the north edge of runway 9. At this time Flying Tiger was 55 feet in the sie and about 2,800 feet from Delta. It would have been physically impossible for the Delta first officer to have seen Flying Tiger until 0910:25 — even if he had been leaning forward in his seat. According to the diagram, Appendix D, Near Collision Tracks, and using an are of 137 degrees (the maximum number of degrees that the first officer could see from the normal cockpit position), he had a view of no more than 575 feet down the centerline of runway 9. Using an absolute reasonable maximum visibility are of 145 degrees, his view along the centerline was approximately 1,960 feet. At this time, Flying Tiger was still 2,405 feet from Delta. With Delta on a heading of 118 degrees, and Flying Tiger on a heading of 90 devrees, the Flying Tiger aircraft woud be beyond the 145 degree are, which is more than 60 degrees behind the Delta first officer's shoulder. This acute angle would have mede a sighting of Flying Tiger beyond the extreme physical limits of visibility from the Delta cockpit. Further, the Board does not discuss the poor judgment of the ground controlier in clearing Delta to cross the runway at 9910:00 when Delta was approximately 800 feet from clearing the south edge of runway 9R. The controller stated he had observed a radar target 3.5 miles from the runway threshold at this time. According to the flight data recorde: , Flying Tiger had an average approach air speed of 190 mph (180 mph ground speed), or 3 miles per minute. Flying Tiger would have been over the runway in 70 seconds, It would have taken Delta at least 60 seconds to taxi the 800 feet to completely clear runway 9. In my opinion, 10 seconds is not a sufficiently safe margin. As a result of this accident and several other runway incursion accidents and incidents, the Board should have recommended to the FAA that either positive coordination be required between ground and local contrel with no exemptions before an aircraft is cleared to cross an active runway, or that only the loced controller should have the authority to issue a taxi clearance to cross an active runway. In conelusion, 1 would not have included Delta as @ primary cause to this accident, because it was physically impossible for tne Delta crew to have seen Flying Tiger until it was too late due to restricted metecrological conditions and physical visual limitations from the Delta cockpit. /s/ FRANCIS H. McADMAS Member a 5. APPENDIXES APPENDIX A Investigation and Hearing L Investigation The National Transportation Safety Board was notified of the accident about 0915 on February il, 1979. Investigators from the Safety Board's Chicago Field Office and Washington, D.C., headquavtes went to the scene. Working groups were established for operations, air traffic control, systems, and structures.

[ANALYSIS] The captain could have seen the same distance only if he had ‘eaned froni his normal eye position, which he said that he cid. The B~-747 was beycnd their visual range. At 0910:19, FT 74, while 55 ft above the runway, was about 2,800 ft from DL 349 and within visual range. As FT 74 descended over the runway, the aircraft may have been obscured initially to ground observers by the fog bank at the end of runwey 9R. Further, the predominant silver-gray fuselage colors of the B-747 would have been inconspicuous against the aull colors of the overcast. The aircraft's landing lights were not illuminated and thus lessened easy recognition of the aireraft's position. At this time, DL 349 was 80 ft from the runway and the aircraft hed turned to a magnetic heading of 18°, At that position, if the first office: ha* moved tis head forward and to the right, whieh would have enlarged his vision ¢..vslope, i: would have been within his capability to see the approaching aircraft .+d warn the captain to Stop the aircraft. However, from the normal eye reference position, the first wi'ficer could not have seen FT 74 until aoout 4 seeonds before the near collision. Although the response time was minimal, it probably was suffieier.t to have stopped the aircraft and permitted PT 74 to pass safely on the runway. The Delta captain could not have seen the B-747 from his position on the left side of his aircraft t. DL 349, at this time, began a turn bac< to the right, reaching a heading of i28° as the B~727 entered the runway at 0916:26. Meanwhile, FT 74 had landed at 0910:24 and was within 1,100 ft of DL 349. The B-747 had now entered the normal vision envelope of the Delta first officer. The first officer could have seen FT 74 if he had icoked in the direction of the runway §K laading area. At this point, the DL 349 captain still could not see FT 74 which was well outsid2 his vision envelope. When the aircraft were about 850 ft apart, FT 74 started to head off the runway and the FT 74 first officer shouted a radio warning for DL 349 to Stop. Because the two aircraft had been assigned different ATC communications frequencies, neither flizhterew heard the ATC clearances issued to the other flight and DL 349 could not near the warning transmitted by the FT “4 first officer. It was not until the 8-747 was within 150 tt that the DL 349 captain could have seen the other sireraft. The Sefety Board notes that 14 CFR 913 holds the pilot in command of an aircraft to be directly responsible for, and the final authority as to, the operation of thst aircraft.

[ANALYSIS] The ARTS II coraputer at ORD is prorrammed to display data tags for transponder-equipped arrivals to a point 2 mi from the approach end of runway 9R. At this point the data tag is automatically dropped and the basic radar target is cinphasized with an asterisk overlay as the taryet continues toward the runway threshold. The lozal controller confirmed his observation of the orogrammed asterisked target of FT 74 while he Ssimritaneously monitored the progress of UA 225 on landing rollout. — tiis obsu’vation of the asterisked radar target when UA 225 was on the runway placed the separation of the two heavy aircraft at less than the required 4 mi. further, his observation confirii3 that the radar target of FT '’4 was displayed and should have been observed by the ground controller. Ai 0919:00, when the outbound ground controller issued the clearance for DL 349 to continue across runway 9R, DL 349 was turned onto the I4R parallel taxiway. Simultaneously, FT 74 was on a collision course from a point 335 ft a.g.L and about 8,000 ft from the in‘ :rsection of the parallel texiway and the landing runway. Under the ORD ARTS fll pregram, FT 74 would have been represented on the tower BRITE display as an asterisked target. The Safety Board believes that when the ground ccntroller looked at the radar, he anticipated that he would observe the arrival flight positioned behind UA 225 to be 4 mi in trail. When the ground gontroller seanned the BRITE display he failed to see the asterisked target associated with FT 74. Since he did not expect the next arrival flight to be sequenced so closely, he aecepted a more conspicuously displayed target, with data tag, as the next arrival. This target represented TW 281, which was about 5.25 nmi behind FT 74. Based on what he perceived to be a 3a.e interval between landing traffic, che ground controller cleared DL 349 to cross runway 9R. The Safety Board conciudes that the ciearanve was the result of the ground controller's incorrect radar target identifieatian. At the time of the accident, visibility at ORD was variable, dependent upon the position of the observer ard the direction of the observation. The 0905 NWS surface weather observation stated that visibility was 1/2 mi, with fog and haze, and also stated that runwsa7 14R visual range was 4,000 ft variable to 4,500 ft. Several minutes befcre the accident, the local controller had advised landing aircraft that runway visual range for runway 9R was 3,900 ft in the touchdown zone and 6,000 ft in the roliout area. He also observ>d the RVR on cunway 9R as 3,000 ft when UA 225 landed although he did not report, as required, the RVR reading to FT 74.

[FINDINGS] Pieces of the wheel rim from either the No. 1 or the No. 2 wheel hit the tire and caused it to blow cut. This blow out affected further the aircraft's braking capability. Also, the left main landing gear might not have collapsed if No. 5 tire had been available to distribute load on the overrun area. The tires on the aircrvaft may have been operaied in the overdeflected condition, since the average inflation pressure was less than the optimum pressure for maximum gross weight. The aircraft left the departure end of runway 6R at a speed of about 68 kns. The aircraft slid to a stop about 83 sec after the start of the takeoff. It came to rest about 664 ft beyond the departure end of runway 6R on a heading of 008°. The airceaft could not be stopped on the available runway because of the partial loss of breking effectiveness attributed to failed tires and a wet runway surface. Dynamic hydiuplaning conditions were not present. Runway 6R had acceptable friction characteristics according to current FAA suggested criteria for the Mu meter; however, the Mu meter data could not be used to estimate aircraft stopping performance. During the 4-year period between the grooving of runway 6R/24L and the day of the accident, the airport operator did not make “he friction surveys suggested by the FAA. The FAA and the airport operators did not have ready access to equipment or trained personnel required to conduct periodic friction surveys. No FAA procedures or data are available to aircraft operators or flightcrew to relate degraded runway friction conditions to changes in allowable aircraft takeoff weights, decision speeds, and stopping distance. -~4l - The current FAA rejected takeoff requirements for aircraft certification, aircraft operations, and pilot training do not address wet ruuway, slippery runway, or tire failure conditions. It was not possible to determine accurately from performance analyses if the full braking capability of the sircraft was achieved during the initial phase of the rejected takeoff. In its 1977 report on rejected takeoffs, the FAA concluded that aircraft safety could be improved by accounting for wet/ slippery runway conditions and tire improvements. Flightcrew simulator training for rejected takeoffs ie inadequate because of the lack of FAA requirements for wet runway considerations in those simulators and for rejected takeoff training at the maximum takeoff gross weights and decision speeds encountered in normal operations. The landing gear attachment structure failed and caused the left wing fuel tank to rupture. Fire may have started before the aircraft left the runway surface. The evacuation was started promptly and elmost simultaneousiy throughout the cabin. The 1L exit was opened with the slide/raft handle in the disarm position. Slide/rafts at exits 2L, 3L, and 4L burned inomediateiy after they were deployed. All slide/rafts on the right side were deployed and used. The overwing ramp for the 3R slide/raft malfunctioned.

[ANALYSIS] Although other runways at the airport, which arg 2,000 ft longer than runway 6R, probably could have contained the rejécted takeoff, they were not available to aircraft with gross weight’ of more than 325,000 lbs because uf runway overpass strength limipations. A project to eliminate this limitation is in the planning stapes. The Safety Board urges the responsiole authcrities to expeaite this project and make longer, safer runways availabl. to heavier aircraft gt Los Angeles International Airport. . i ~ 7 - jf Even though te measured wet friction characteristics of , wunway 6R exceeded minimum ctandards suggested by the FAA, the Safety eens, ; TTL, » -~ 33- Board believes these characteristics contributed to the partial loss of the aircraft's brakirg capabilities and, therefore, contributed to the inabi’ity to stop the aircraft on the runway. This loss of runway friction was particuiar!y evident in the rubber coated areas on the departure end of the runway, the touchdown area for landings on runway 24L. The FAA developed the minimum runway friction standards and methods for the measurement of these ¢tandards. FAA Advisory Circular AC 150/5320-12, Methods for the Design, Construction and Maintenance o° Skid Resistant Airport Pavement Surface:, made this information available to airport operators. Because the inforration in this Advisory Circular is not mandatory, airport operators Jo not routinely use it. At the time of this accident, neither the Los Angeles International Airport cperat>r nor the FAA authorities in the Los Angeles area had the FAArecommended equipment to make these measurements, Furthermore, no record could be found to show that friction surveys had ever been conducted or that rubber deposits and other contaminents had ever been removed from the surface of runway 68/24L since the runway was grooved tn 1974. When the Safety Board made the results of its runway friction tests available, the affected areas were cleaned. For some time the Safety Board has maintained chat the provistons of AC 150/5320-12 should be made mandatory. As a result, the Safety ‘Board issued safety reconmendations A-76-136 and 137 on November 18, 1976. The FAA disagrees with the Safety Board, in that it believes the economic burden placed on individual airport operators would be prohibitive and the precision techniques for friction testing are not presently available. However, they are presently studying the matter. Certification and operations regulations do not take into account the longer stopping distances required by rejected takeoffs on wet or slippery runway surfaces or the reasons for rejected takeoffs other than an engine failure. The FAA reached this same conclusion in its 1977 Jet Transport Rejected Takeoff Study.

[ANALYSIS] The Bresg weight and c.g. were within prescribed limits. The aircraft's airframe systems, and powerplants were not causal to this accident. Tne evidence showed that the accident was initiated by the nearly simultaneous carcass failures of the two tires mouated in the No. 1 and No. 2 positions. Since these tires were mounted on the same axle, che 97,920-ib load on the axle was distributed between the two tires. The analysis of tre and wheel marks on the runway indicated that the failure sequence began when the tread from the No. 2 tire separated from its carcass about 6,300 ft from the departure end of the runway. The tire carcass remained intact until the aircraft was about 4,520 ft from the runway departure end where squiggle marks indicated blowout. The squiggle marks on the runway at that point and portaccident examination of the tire remains indicated that extreme heat had built up in the carcass sidewall and that the carcass had blown out at its upper sidewall. After the treau had separated, the rubber was ~braded by direct contact with the runway surface and eventually blew out. After the No. 2 tire carcass blew out, the entire load on the axle was imposed upon the No. 1 tire. The markings made by the Nv. 1 rim, which showed contact with the runway surface 4,480 ft from the runway deperture end, indicated that the No. 1 tire failed aimost immediateiy (within two wheel revolutions) after the No. 2 tire carcass failed. The No. 1 tire virtually disintegrated while the whole No. 2 tire carcass, except the beads, came off the wheel. Examination of the remains of the No. 1 tire indicated that the tire ultimately blew out in the lower sidewall. The DFDR showed that the tires failed just before the aircraft accelerated through 152 kns--about 4 kns below the calculated V1 speed. The DFDR further showed that the captain reacted promptly to the tire failures and began rejected takeoff procedures. However, he was not able to stop the aircraft within the remaining runway. Thus, to understand this accident sequence, two distinct, but related, issues must be analyzed. First, since the tire failures triggered the sequence of events, tire failuree and tire reliability in general must be anaylzed. Second, reasons must be determined for the captain's inability to stop the aircraft on the runway even though the rejected takeoff was -nitiated before the aircraft reached V, speed. ‘Tire VaiJures and Tire Reliability Both the No. 1 and No. 2 tires were on their third retread cycle, a limit which was wet hv the airline based upon prior experience of unscheduled remova}} of DC-i0-lv tices. The two tires had been manufactured by differen: companies and had different design characteristics. Both, however, met all specifications sec forth in FAA regulation for certification.

[FINDINGS] Their imacdiate response and their initiative in seeking al.ernate escape routes when the normal routes were rendered useless, undoubtedly saved lives and decreased the number of injuries. 3. CONCLUSIONS Findings The crewmembers were certificated and qualified for the flight. The aircraft was certificated, equipped, and maintained in accordance with FAA requirements, except for the inoperative CVR. The runway was vet, but there was no standing water. Runway 6R was the only runway available for takeoff. Two 12,000-ft runways, the use of which could have made a successful rejected takeoff possible, were not available to wide body aircraft. Lineup for takeoff began about 166 ft, from the approach end of runway 6R. The flightcrew used the minimim lineup distance and established takeoff thrust as required by company procedures. The captain promptly rejecced the takeoff at or below 152 kns (V1 speed was 156 kns) after hearing a loud "metallic bang" and feeling a “quivering" of the aircraft. The captain responded to the emergency by first applying brakes and then applying maximum reverse thrust on all engines. Ground spoilers actuated whei thrust levers were moved to the reverse thrust positions. Reverse thrust began about 5.8 sec after V1 was reached and peaked 3 to 8 sec after the engines began to spool up for reverse thrust. Reverse thrust was maintained above 100 percent Ny on ali three engines during the reversal sequence. Reverse thrust was maintained on the center and the right engine until just before the aircraft stopped beyond the end of the runway. Reverse thrust on she left engine ceased when that engine was torn from the aircraft, 100 ft beyond the end of the runway. The first tire feiled at the No. 2 tire position about 6,300 Ft from the departure end of runway 6R. The tire failed because of a thrown tread. The carcass blew about 4,520 ft from the departure end of the runway. The second tire failed at the No. 1 tire position about 4,480 ft from the departure end of rvnway 6R. Fatigue in the ply structure may have been caused ty long-term overload since the tire wis mounted on an axle with a tire of a different brand whic) had less sfdewall stiffness. The tire blew out because of an overload. The third tire failed at No. 5 tire position about 3,400 ft from the departur2 end of runway 62. Pieces of the wheel rim from either the No. 1 or the No. 2 wheel hit the tire and caused it to blow cut. This blow out affected further the aircraft's braking capability. Also, the left main landing gear might not have collapsed if No. 5 tire had been available to distribute load on the overrun area.

[ANALYSIS] However, they are presently studying the matter. Certification and operations regulations do not take into account the longer stopping distances required by rejected takeoffs on wet or slippery runway surfaces or the reasons for rejected takeoffs other than an engine failure. The FAA reached this same conclusion in its 1977 Jet Transport Rejected Takeoff Study. However, the FAA still he3 not developed ‘rocedures which would allow aircraft manufacturers, airline operators, or flight crewmembers to determine changes in decision speeds or aircraft gross weights, or both, so that successful rejected takeoffs could be accomplished from near V1 peed on a wet runway following engine or tire failures. The Safety Board c ...curs with the recommendations in the FAA's study and believes that, unless FAA takes wet or slippery runways and the other reasons for rejected takeoffs into account, accidents and incidents will continue, especially when aircraft are required to operate on dry, wet, or slippery runways of critical length. Flightcrews, for the most part, are trained for rejected takeoffs ta flight simulators, and vherefore, training is limited by the capability of the simulator and by the training requirements of 14 CFR 121. Since the simnulator's accelerate-stop performance is based on the aircraft / / cA manufacturer's dry runway, engine-out fertification data, it is impossible to simulate realistic wet runway conditions or malfunctions other than an engine failure. FAA simulator rgquirements contain performance specif{{icaticns for the acceleratios{{ portion of the rejected takeoff maneuver, but nct for the stopping portion of the waneuver. Further, the FAA does not requfre that eye instructor, evaluation pilot, or trainee determine pilct reactich time and the amount of braking effort ‘applied by pilots in the simusator. The training reqy rements of 14 CFR 121 do not require rejected takeoffs at the maximum gross weights and decision speeds encountered in normal operations. As revealed in testimony at the public hearing, crewmembers do not typigally receive this more demanding training. Although a captain is expected to know that a maximum braking effort would be required when a rejected takeoff is initiated at the higher speeds and gross wejghts, he cannot be expected to judge whether the aircraft is decelerating at its maximum capability if he has never been trained for that eventuality. The captain of Flight 603 reacted promptly to the tire failures, and he acted iff accordance with Continental procedures. During depositions, he stated that he applied full brake pressure immediately. However, because there was no requirement for the DFDR to record brake pressure at the brake or flight test data available to validate the Douglas/NASA predicted deceleration rates, the Safety Board was not able to verify if maximun brake pressure was achieved during the early porcion of the rejected takeoff. Further, the Safety Board could not determine if the antiekid system performed to its maximum capability during the sane time period.

[ANALYSIS] The onset of this wave depends on the ground-speed of the aircraft. -~32- In this instance, tire No. 2 weuld have then been overloadec. and overdefiected which could result in tread loss followed by sarcass blowout. Failure of the No. ? tire alone probably would not have affected the aircraft's accelerate-st » performance to the extent that an accident was inevitable. The Safety Board believes that had the No. 1 tire not euffered previous degradation, it would have been capable of operating for a longer period than evident in this accident. The examination of the tire's carcass disclosed advanced fatigue in the ply structure. In addition, there was evidence of severe cord overheating near the sidewall vead area, ari several other areas of the sidewall showed evidence of very nigh temperatures. Such corditions are typical of those produced by overload or overdeflected operation for a prolonged period of time. Although the Safety Board cannot determine when such damage was inflicted, ft is concerned that airframe and tire design, aad operational and maintenance procedures can combine to cause prolonged operation of tires In an overdefiected or overloaded condition. Normal differences between two tires on the same axle, particularly if they are of different designs, could preclude them from carrying equal loads. The Safety Board believes that the preexisting damage in the No. 1 tire was a facthr in causing it to ultimately fail almost immediately after the No. 2 tire failed, and thus, the preexisting damage may have been a causal taftor. Abbut 3,400 ft from the departure end of runway 6R, the No. 5 tire fail:43. "This failure was caused by foreign object damage when pieces of either the No. 1 wheei or No. 2 wheel broke off after the wheels contacted the runway surface and hit the No. 5 tire. This failure further reduced the braking capability of the aircraft. wv Rejected Takeoffs == “=> a Because of its gross weight of about 430,900 lbs, the only runway available to the aircraft at Los Angeles International Airport wae 4R, which was 10,285 ft long. Pased on current FAA dry runway certivication data, an 850-ft stopping margin would be expected if a reject.:1 ta off was initiated at Vy because of engine failure. However, wher et rutiway surface conditions and tire failures are considered, the stopping margin is eliminated. Although other runways at the airport, which arg 2,000 ft longer than runway 6R, probably could have contained the rejécted takeoff, they were not available to aircraft with gross weight’ of more than 325,000 lbs because uf runway overpass strength limipations. A project to eliminate this limitation is in the planning stapes.

[ANALYSIS] Furthermore, since the aircraft's performance data were obtained through testi ig on dry runways, there is no assurance that the current concept is adequate when the braking coefficient of friction is reduced on a wet surface. Even when full braking capability js available and the runway surface is dry, a rejected takeoff initiated at or just before the aircraft reaches Vj speed is risky on a minimum length runway. Using maximur braking and optimum procedures, the aircraft is going to use all of the remaining runway length to stop. (Actually, a smail margin is provided since the braking effect of thrust reveraal is not considered in acceleratestop performance data.) In this accident, the aircratt had about 800 feet more than the minimum runway provided by the balanced field concept; even so. ‘wo significant factcrs combined to invalidate the use of V1. speed €3 .. , »-no-go decision point: (1) the loss of effective braking on whee]: 7:th blown tires, and (2) the reduction in brake friction coefficient on wet surface. The Safety Borrd, therefore, views the captain's no-go decision as a key element in the accident sequence. This is in no way intended to imply that the Board faults his decision, but rather that the limited validity of the decision-making process and of the V}} concept in its entirety justifies further analysis. As the aircraft approaches the decision apered (V1) the decisionmaking time available to the pilot decreases. At Vj speed he has no time for a decision and must respond immediately to reject the takeoff {{f he is to be able to stop the aircraft on the runway even under ideal conditions. A failure to act promptly to any problem encountered as the aircraft approaches the Vj speed can have catastrophic results if the problem is an engine failure, is structural in nature, or is associated with loss of critical systems or fiight controls. Certainly, these possibilities are ever present in a pilot's mind when something unusual occurs at a critical time. Therefore, the dominant tendency is most likely to reject the takeoff when any unevaluated anomaly occurs before Vi speed, particularly if the problem is accompanied by noise and vibration. In this accident, the captain heard a loud metallic bang and the flight data recorder indicated that this occurred 1.2 seconds before the aircraft reached Vj speed. The captain was therefore faced with the need for immediate action. He had no time in which to evaluate the significance of the loud bang and vibration if he was to successfull: reject the takeoff. However, it became evident during the Board's investigation that the noise and vibration were associated with a tire failure and that the aircraft could undoubtedly have been flown cff the runway successfully.

[ANALYSIS] However, because there was no requirement for the DFDR to record brake pressure at the brake or flight test data available to validate the Douglas/NASA predicted deceleration rates, the Safety Board was not able to verify if maximun brake pressure was achieved during the early porcion of the rejected takeoff. Further, the Safety Board could not determine if the antiekid system performed to its maximum capability during the sane time period. After the captain and the other flight crewmembers became aware that they would not te able to stop the aircraft on the runway, the captain steered the aircraft to the right to avoid colliding with the approach light stanchions fcr runway 24L. The Safety Board believes that this action reduced the severity of this accident. Impact with the nonfrangible stanchions could have caused additional major structural © damage. The Rejected Takeoff Decision The determination of the minimum length of runway required for takeoff in air carrier operations, or conversely the determination of the maximum weight for the airplane to take off on any given runway, is based upon a balanced field concept. This concepe is predicated upon the calculated ability of the aircraft to either stop within the length of the runway or to successfully continue the takeoff after an engine failure during the takeoff roll. Before euch takcoff, the flightcrew will use accelerate-stop performance data obtained from the aircraft's certification tests to calculate the maximum allowable takeoff weight and the critical engine fullure speed (V)). The flightcrew has been trained to use the Vy speed as a decision point during the takeoff roll. If au engine failure ie recognized before the V) speed is reached, the pilot is trained to reject the takeoff and ne, in fact, must reject tne takeoff since he cannot be assured of successfully continuing. On the other hand, $f the aircraft is beyon?’ the V1 speed before ar engine failure is recognized, the takeoff must be centinued since the pilot cannot be assured of stopping the aircraft on the remaining runway. Although V) speed is designed to be che go-no-go decision speed in event of an engine failure, the Safety Beard believes that pilots have come to regard Vj}} as the go-no-go decision speed for any anomaly during the takeoff roll. However, the calculated Vj speed, by current definition and certification standards, is valid only fer circumstances in which the aircraft has its full braking capa>ility. Furthermore, since the aircraft's performance data were obtained through testi ig on dry runways, there is no assurance that the current concept is adequate when the braking coefficient of friction is reduced on a wet surface. Even when full braking capability js available and the runway surface is dry, a rejected takeoff initiated at or just before the aircraft reaches Vj speed is risky on a minimum length runway.

[ANALYSIS] The tire marks over the next 200 feet of travel indicated that the airplane departed the runway at an angle of 10.4o to the left of the runway centerline. 40 2. ANALYSIS 2.1 General The flightcrew was properly certificated and qualified in accordance with applicable regulations and company requirements. All three crewmembers were experienced at their respective positions. Evidence from crew duty time, flight time, rest time, or off-duty activity patterns did not indicate that behavioral or physiological factors affected the flightcrew on the day of the accident. The ATC personnel involved with the flight were all properly certificated and qualified. The airplane was properly certificated, equipped, and maintained (with the exception of the FDR system) in accordance with FARs and approved regulations. The weight and balance were within allowable limits. In analyzing this accident, the Safety Board focused on flightcrew actions and decisions, B-747 procedures for slippery runway operations, the performance of air carrier training simulators for B-747 operations on slippery runways, flight attendant actions and cabin safety issues, Tower Air management oversight of maintenance and operations, FAA surveillance of Tower Air, and FAA policies and procedures regarding the evaluation of slippery runways. 2.2 Flightcrew Actions and Decisions 2.2.1 Pre-takeoff Events Although the flightcrew was not provided the runway friction values obtained by the airport operations crew, they had obtained sufficient indications from the slipperiness of the taxiways, the appearance of runway 4L, and the blowing snow to recognize that they were operating in a challenging environment of wind, reduced visibility, and runway slipperiness. Based on the existing surface and wind conditions on the day of the accident, the captain might have considered using runway 31L (which was more favorably oriented to the wind) for his departure. However, when the captain overheard the response of JFK ground control to another flight’s inquiry about runway 31L that it would remain closed for another couple of hours, he determined that runway 31L was not a viable option for departure. Although 5 minutes before the accident ATC changed the departure runway to 31L for traffic following flight 41, the Safety Board recognizes that the captain’s decision to use runway 4L was based on the limited information available to him at the time. Further, air traffic controllers were not required to offer flight 41 the option of switching to runway 31L, once the airplane was established holding short at runway 4L.

[ANALYSIS] Further, there are no means to compare measurement standards or translate the data into aircraft performance. A key issue is that no significant progress has been made in correlating stopping distance data from airplane manufacturers’ flight tests and calculations with the friction values obtained from measuring devices. An outcome of these correlations could be the establishment of objective standards for air carrier operations on slippery runways, perhaps extending to the establishment of appropriate minimum runway friction levels for operational use. The Safety Board concludes that the circumstances of this accident indicate that the issue of correlating airplane stopping performance with runway friction measurements should be revisited by the Government and the air transportation industry. Consequently, the Safety Board believes that the FAA should require the appropriate Aviation Rulemaking and Advisory Committee to establish runway friction measurements that are operationally meaningful to pilots and air carriers for their slippery runway operations (including a table correlating friction values measured by various types of industry equipment), and minimum coefficient of friction levels for specific airplane types below which airplane operations will be suspended. 55 3. CONCLUSIONS 3.1 Findings 1. The flightcrew was properly certificated and qualified in accordance with applicable regulations and company requirements. 2. The air traffic control personnel involved with the flight were all properly certificated and qualified. 3. The airplane was properly certificated, equipped, and maintained (with the exception of the flight data recorder system) in accordance with approved regulations. The weight and balance were within allowable limits. 4. The captain’s decision to attempt the takeoff on runway 4L was appropriate. 5. Asymmetric thrust was not a factor in the loss of directional control. 6. The captain’s failure to correct the airplane’s deviation from the centerline resulted from his overcontrolling the nosewheel steering through the tiller. 7. The captain of flight 41 first relied on right tiller inputs as the airplane continued to veer left, then applied insufficient or untimely right rudder inputs to effect a recovery. 8. Current Boeing 747 operating procedures provide inadequate guidance to flightcrews regarding the potential for loss of directional control at low speeds on slippery runways with the use of the tiller. 9. The procedural change by Tower Air to reevaluate and eliminate its standard procedure of guarding the tiller during the takeoff roll through 80 knots will make overcontrol of the tiller less likely for its own operations; however, other air carrier operators of the Boeing 747 may need to make similar changes to their procedures. 10. Current Boeing 747 flight manual guidance is inadequate about when a pilot should reject a takeoff following some indication of a lack of directional control response. 11. Improvements in the slippery runway handling fidelity of flight simulators used for Boeing 747 pilot training are both needed and feasible. 12. The captain’s failure to reject the takeoff in a timely manner was causal to the accident. 13. The inadequate Boeing 747 slippery runway operating procedures developed by Tower Air and the Boeing Commercial Airplane Group, and the inadequate fidelity of B-747 56 flight training simulators for slippery runway operations, contributed to the cause of this accident. 14.

[ANALYSIS] Although 5 minutes before the accident ATC changed the departure runway to 31L for traffic following flight 41, the Safety Board recognizes that the captain’s decision to use runway 4L was based on the limited information available to him at the time. Further, air traffic controllers were not required to offer flight 41 the option of switching to runway 31L, once the airplane was established holding short at runway 4L. Based on the absence of definitive runway friction measurements for runway 4L, reported winds of less than 15 knots (the maximum recommended crosswind component for B-747 takeoffs on slippery runways), the flightcrew’s reports of acceptable visibility down the runway, and the reported unavailability of the alternative runway 41 31L, the Safety Board concludes that the captain’s decision to attempt the takeoff on runway 4L was appropriate. 2.2.2 The Attempted Takeoff and Loss of Control Flight 41 attempted its takeoff under crosswind conditions with a runway contaminated with packed snow and patchy ice. At the approximate time of the takeoff attempt, there were crosswinds of 10-12 knots. Gusts of up to 22 knots were reported in the general area near the time of the accident. Asymmetric thrust (for example, inadequate thrust from the No. 1 engine) could have resulted in the loss of directional control experienced by flight 41. In the absence of a cross-check of other engine instruments, a malfunctioning EPR indicator could have led the flightcrew to unknowingly set inadequate thrust for the No. 1 engine. However, given the flight engineer’s recollections of evenly matched engine acceleration and consistent EPR and N1 indications from all four engines, and the absence from the CVR of flightcrew discussions of abnormal throttle alignment, the Safety Board concludes that asymmetric thrust was not a factor in the loss of directional control. Having verified the realism of the Boeing engineering flight simulator in reproducing the ground handling characteristics of the B-747 on slippery runways, the Safety Board applied the findings of its August 8, 1996, flight simulation study to the circumstances and events in this accident. In all simulations in which the pilot did not use the nosewheel steering tiller for directional control (including those conducted with icy runway conditions and winds gusting up to 40 knots), the simulated airplane was controllable along the runway centerline. In contrast, when pilots attempted to maintain the runway centerline using the tiller under slippery runway conditions with a 12-knot crosswind, a slight overcontrol at the very beginning of the takeoff roll repeatedly led to the loss of traction and steering capability from the nosewheel, followed by the loss of directional control. Given that it is very unlikely that the captain did not try to control the airplane’s tendency to weathervane into the crosswind, and given the consistent controllability of the airplane under accident conditions when the tiller was not used (during the simulation study), the Safety Board concludes that the captain’s failure to correct the airplane’s deviation from the centerline resulted from his overcontrolling the nosewheel steering through the tiller.

[ANALYSIS] Although the airport personnel claimed that the report was made, there was no documentation of a timely report in their records; the only such record was of a postaccident entry in the operations office computer. The control tower was required by FAA Order 7110.65J to advise pilots of runway friction readings when they were received from airport management, but the control tower personnel claimed that they did not receive these reports. The Safety Board 27 Federal Aviation Administration. FAA 90 Day Safety Review. Washington, DC. September 16, 1996 (mimeo). 54 was unable to determine whether the runway friction measurement data were sent or received. However, the Safety Board concludes that the failure of the PNY&NJ or FAA air traffic control tower personnel to provide these data to the pilots of flight 41 did not contribute to this accident. Although the guidance currently provided by the FAA on runway friction measurement and reporting may be helpful to airport operators, it is incomplete because friction coefficient measurements of various types are not correlated with braking performance of different airplane types or configurations. The International Civil Aviation Organization (ICAO) Guidance Material Supplementary to Annex 14, Volume I, 6, includes a table of friction coefficient measurements correlated with descriptive values, i.e., good, medium, poor. However, this table is provided for informational use only, and it, too, does not establish clearly defined parameters applicable to airplane types. The Safety Board is concerned about the frequent occurrence of veeroffs, overruns, and other related events by large airplanes when runways are contaminated with ice, snow, and/or slush (including this accident). The continuing problem with safety during ground operations is related to several problems. There is a clear need to measure the slipperiness of runway and taxiway surfaces. However, those values must then be quantified into meaningful information that pilots can use to evaluate the expected performance of their specific airplane. This would require airport operators to maintain their equipment within specific tolerances, and it would require the technicians operating the equipment to adhere to appropriate standards in using the equipment. If the FAA had been responsive to the Safety Board’s 1982 safety recommendations on this subject (see section 1.10.3), the industry might have already resolved these problems. The FAA has made considerable progress in providing and implementing procedures for airport operators to perform friction measurements during periods of ice/snow and slush contamination. However, such measurements are still not required, and there is no standardization of the equipment currently being used. Further, there are no means to compare measurement standards or translate the data into aircraft performance. A key issue is that no significant progress has been made in correlating stopping distance data from airplane manufacturers’ flight tests and calculations with the friction values obtained from measuring devices. An outcome of these correlations could be the establishment of objective standards for air carrier operations on slippery runways, perhaps extending to the establishment of appropriate minimum runway friction levels for operational use.

[PROBABLE CAUSE] Inadequate Boeing 747 slippery runway operating procedures developed by Tower Air, Inc., and the Boeing Commercial Airplane Group and the inadequate fidelity of B-747 flight training simulators for slippery runway operations contributed to the cause of this accident. The captain’s reapplication of forward thrust before the airplane departed the left side of the runway contributed to the severity of the runway excursion and damage to the airplane. 58 4. RECOMMENDATIONS As a result of the investigation of this accident, the National Transportation Safety Board makes the following recommendations: --to the Federal Aviation Administration: Require modification of applicable operating procedures published by the Boeing Commercial Airplane Group and air carrier operators of the B-747 to further caution flightcrews against use of the tiller during slippery runway operations, including low-speed operations (for airplanes equipped with rudder pedal steering) and to provide appropriate limitations on tiller use during these operations (for airplanes not equipped with rudder pedal steering). (A-96-150) Issue a flight standards information bulletin to principal operations inspectors assigned to air carriers operating the B-747, informing them of the circumstances of this accident and requesting a review and modification, as required, of each air carrier’s takeoff procedure regarding pilot hand position with respect to the tiller. (A-96-151) Require the Boeing Commercial Airplane Group to develop operationally useful criteria for making a rapid and accurate decision to reject a takeoff under slippery runway conditions; then require that B-747 aircraft flight manuals, operating manuals, and training manuals be revised accordingly. (A-96-152) Evaluate Boeing 747 simulator ground handling models and obtain additional ground handling data, as required, to ensure that B-747 flight simulators used for air carrier flightcrew training accurately simulate the slippery runway handling characteristics of the airplane. (A-96-153) After completing this evaluation, issue a flight standards information bulletin urging principal operations inspectors assigned to air carrier operators of the Boeing 747 to enhance simulator training for slippery runway operations, including limitations on tiller use and instructions for rudder use during the takeoff roll. (A-96-154) Develop certification standards for the installation of secondary galley latches; then use those standards to conduct an engineering review of secondary galley latches on all transport-category 59 aircraft.

[ANALYSIS] The Safety Board concludes that this procedural change by Tower Air will make overcontrol of the tiller less likely for its own operations; however, other air carrier operators of the B-747 may need to make similar changes to their procedures. Consequently, the Safety Board believes that the FAA should issue a flight standards information bulletin (FSIB) to POIs assigned to air carriers operating the B-747, informing them of the circumstances of this accident and requesting a review and modification, as required, of each air carrier’s takeoff procedure regarding pilot hand position with respect to the tiller. The Safety Board recognizes that it may be a natural reaction for a pilot to persevere in a takeoff attempt when faced with an apparently minor hesitation of an airplane to respond to rudder input. However, the circumstances of this accident indicate that during takeoff in a B-747 on a slippery runway, the pilot must abort at the very first indication of a directional control loss. The Boeing B-747 Operations Manual and Tower Air B-747 Flight Manual direct pilots who are performing takeoffs on slippery runways to immediately reject the takeoff if deviations from the runway centerline cannot be controlled. While this accident demonstrates the soundness of this advice, the accident also indicates that the provisions in these manuals are not adequately specific, particularly in their references to deviations that “cannot be controlled.” Tower Air’s chief of flight standards suggested a criterion for rejecting takeoffs under slippery runway/crosswind conditions that may be useful for pilot decisionmaking in the future. He linked the takeoff rejection decision to the recommended procedure of limiting rudder pedal steering input to one-half full travel to get optimal cornering friction. He indicated it was clear that if a pilot could not control the airplane with one-half rudder pedal travel, the takeoff should be rejected. This advice may be operationally useful for all B-747 pilots, if it can be verified by the FAA and aircraft manufacturer. The Safety Board concludes that current B-747 flight manual guidance is inadequate about when a pilot should reject a takeoff following some indication of a lack of directional control response. Consequently, the Safety Board believes that 45 the FAA should require Boeing to develop operationally useful criteria for making a rapid and accurate decision to reject a takeoff under slippery runway conditions; then require that B-747 aircraft flight manuals, operating manuals, and training manuals be revised accordingly. 2.2.5 Training Simulators for B-747 Slippery Runway Operations The air carrier and FAA pilots who participated in the August 8, 1996, simulation study believed that the Boeing engineering simulator had more realistic ground handling performance than the simulators Tower had provided for pilot training.

[ANALYSIS] The Board is concerned that air carrier B-747 pilots currently are not able to obtain needed training on slippery runway procedures, including proper tiller and rudder techniques, because training simulators have not incorporated the latest ground handling model (such as that implemented on the Boeing engineering simulator). Further, although existing flight test data on slippery runway handling characteristics are limited, the increasing use of high capacity FDRs and quick access maintenance recorders enables data on slippery runway handling to be obtained from actual line flying experience. Many B-747-400 models are equipped with these recorders. The Safety Board concludes that improvements in the slippery runway handling fidelity of flight simulators used for B-747 pilot training are both needed and feasible. Consequently, the Safety Board believes that the FAA should evaluate B-747 simulator ground handling models and obtain additional ground handling data, as required, to ensure that B-747 flight simulators used for air carrier flightcrew training accurately simulate the slippery runway handling characteristics of the airplane. The Safety Board also believes that after completing this evaluation, the FAA should issue an FSIB urging POIs assigned to air carrier operators of the B747 to enhance simulator training for slippery runway operations, including limitations on tiller use and instructions for rudder use during the takeoff roll. 2.2.6 Summary of Flightcrew Actions and Decisions The captain’s use of the tiller control for nosewheel steering during the takeoff roll, combined with his untimely or inadequate use of rudder inputs, allowed the loss of directional control to develop. As this occurred, the airplane’s deviation from the centerline and its unresponsiveness to steering inputs provided cues that, regardless of the adequacy of existing procedures and training methods, should have prompted the captain to reject the takeoff more quickly than he did. Therefore, the Safety Board concludes that the captain’s failure to reject the takeoff in a timely manner was causal to the accident. Still, better procedures for operating the B-747 under slippery runway conditions and improved ground handling fidelity of the flight simulators used for B-747 pilot training could have better prepared the captain for handling the situation that confronted the accident flight.

[ANALYSIS] Although the Safety Board’s fuel calculations do not demonstrate that the accident airplane was completely out of fuel, the Safety Board considered a critically low fuel state to be the most likely scenario given the lack of evidence indicating mechanical failures. Federal regulations and company policy state that it is the pilot-in-command’s responsibility to ensure before departure that the airplane has the correct amount of fuel onboard for the planned flight. The Safety Board concludes that because of miscommunication between the flight crew and the 40 Banking and acceleration can cause fuel to flow away, or become unported from, the airplane’s fuel pumps, which are located in the inboard section of the fuel tanks near the fuselage. 34 NTSB/AAB-02/05 fueler about the amount of fuel to add at FRG and the flight crew’s failure to ensure an adequate fuel supply, the airplane departed FRG with less fuel than the flight crewmembers stated on the load manifest and with less fuel than they believed they had. Further, the Safety Board concludes that the flight crewmembers failed to adequately monitor the airplane’s fuel state en route and at ACY based on the expectation that they had sufficient fuel on board to complete both flights. PROBABLE CAUSE The National Transportation Safety Board determines that the probable cause of this accident was the flight crew’s failure to ensure an adequate fuel supply for the flight, which led to the stoppage of the right engine due to fuel exhaustion and the intermittent stoppage of the left engine due to fuel starvation. Contributing to the accident were the flight crew’s failure to monitor the airplane’s fuel state and the flight crew’s failure to maintain directional control after the initial engine stoppage. Adopted on August 26, 2002

[ANALYSIS] Fuel burn is also increased during rapid, highperformance climbs, such as required for the execution of a missed approach. It is also possible that fuel load information for the accident airplane documented in the days before the accident did not accurately represent the airplane’s fuel load during this period. Carry-over errors in fuel load documents and differences in fuel calculations could result in assumptions of more fuel than was actually on board the airplane. For example, the Safety Board notes that its fuel calculations were based on the airplane being loaded to its maximum fuel load capacity, an amount that flight crews stated was rarely loaded. Using a 2,900-pound top-off fuel amount38 instead of the maximum 3,090 pounds (as assumed in the fuel burn calculations) for these instances resulted in only 85 pounds of fuel remaining. Assuming that as much as 66 pounds were unusable, the remaining 19 pounds, or about 3 gallons, would have been enough for only about 1 to 4 minutes of flight. Therefore, assuming that these factors account for some or all of the fuel that would have remained under the conditions assumed for the high-speed cruise setting calculations and given the larger amount of fuel loaded in the left fuel tank, the right engine stoppage could be explained by fuel exhaustion.39 However, the Safety Board notes that 36 In a postaccident interview, the owner and CEO of Executive Airlines stated that the airlines’ pilots could determine current fuel loads by subtracting fuel burned from the known quantity at departure and that he believed that the fuel burned gauges are more reliable than the fuel quantity gauges. 37 Although Executive’s owner and CEO stated that flight crew used long range cruise settings, the Safety Board considered the airline’s director of operations to be more credible based on his direct interaction with line pilots as their supervisor and as a Jetstream 31 check captain and instructor. 38 As previously noted, Executive Airlines flight crews indicated that it was common practice to list 2,900 pounds on the load manifest because tanks are rarely filled to the top. 39 Witness testimony indicated that the right tank contained less fuel when the airplane departed FRG. The airplane’s fuel crossfeed valve was found in the closed position. However, it was not possible to determine whether this valve was open before the accident (as is required by the airplane checklist in the event of an engine failure). 33 NTSB/AAB-02/05 both the 275-pound and 85-pound flight time estimates indicate a critically low fuel state, and a fuel state well below the reserve required by the IFR flight plan filed by the flight crew. Further, the apparent difficulties with the left engine could be explained by unporting of whatever fuel remained in the left tank.40 Conditions would have been conducive to unporting if, for example, the airplane was in a left bank and in a nose-right sideslip following the loss of the right engine, which is consistent with flying straight while attempting to turn to the left, as indicated by the pilot statement “we’re trying” when the controller queried if the airplane was in a left turn.

[ANALYSIS] Assuming some usable fuel was in the left tank, it is probable that the flight crew’s maneuvering after the right engine stoppage may have intermittently unported it, causing the sequential stoppage and restarting of the left engine. The Safety Board notes that the loss of the right engine occurred while the airplane was in relatively level flight and that the left engine stoppage occurred during attempts to maneuver back to the ILS course after the stoppage of the right engine. Further, the final loss of control may have been precipitated by momentary stopping or starting of the left engine when the airplane was at a reduced airspeed. Impact witness marks and airplane damage were consistent with a loss of control at low altitude. Observations made during examination of the accident site and the surrounding area were consistent with only a small amount of fuel being on board at the time of the crash. Severe fire damage was limited to the area immediately surrounding the crash site, suggesting a lack of significant fuel splash from the ruptured tanks. The burn pattern also indicated that there was little fuel remaining on board at impact. Fuel starvation to the engines caused by the failure of the LP fuel cock circuitry was also considered. However, no determination could be made because the LP fuel cock switches located in the cockpit were destroyed by impact and fire. The Safety Board notes that the right engine and left engine LP valves, which are controlled by the fuel cock circuitry, were found in the open position and that the hydraulic LP valves were found open. Loss of fuel en route was also considered. However, witnesses who observed the airplane’s departure from ACY saw no indication of leaking fuel from the wings or the fuel caps. Although the right tank fuel cap was not recovered, metallurgical evidence determined that it had been in place before impact. Further, the airplane’s manufacturer indicated that, because of the outboard location of the fuel caps, spillage or suction from an open fuel tank filler in level flight would be minimal and unlikely. Although the Safety Board’s fuel calculations do not demonstrate that the accident airplane was completely out of fuel, the Safety Board considered a critically low fuel state to be the most likely scenario given the lack of evidence indicating mechanical failures. Federal regulations and company policy state that it is the pilot-in-command’s responsibility to ensure before departure that the airplane has the correct amount of fuel onboard for the planned flight.

[ANALYSIS] Further, the Safety Board notes that during simulator tests, when a left engine failure was induced along with a failed right engine on an ILS approach, the airplane remained controllable with minimal difficulty. However, the airplane was more difficult to control and required constant flight control and trim inputs when the left engine was intermittently failed and restored.33 Because of impact damage to the empennage, and the stretching and twisting of the rudder trim control cables, an exact rudder trim position for the accident airplane could not be determined. The Safety Board also considered the possibility of a propeller governor malfunction. However, postaccident examinations determined that both propeller governors were functional. Further, had an internal governor failure occurred (such as a seized drive shaft), the loss of oil to the propeller dome would have caused the feathering spring to drive the propeller into feather and would not have caused significant control problems. Airplane Fuel Loading Three witness accounts provided information about how much fuel was on board the airplane at FRG before fuel was added in preparation for the first flight leg of the day to ACY: • The captain who cancelled the flight earlier in the day because of weather and because it was too late at FRG for the airplane to be refueled stated that the airplane had 1,200 pounds of fuel on board; • The Executive Airlines owner and CEO stated that the accident first officer told him before departure that the fuel tank measuring sticks indicated that 1,100 pounds of fuel were on board; and • The first officer’s fiancée, who was a passenger to ACY, stated that the first officer indicated that 1,200 pounds of fuel were on board and told her that one fuel gauge read 300 pounds and that the other read 900 pounds. She stated that she believed that the 300-pound indication was for the right tank. These reports are roughly consistent with Safety Board calculations, based on airplane log pages and flight crew records (between May 19 and before the accident airplane’s fueling in FRG), which indicated that about 1,000 pounds of fuel were onboard the airplane before the 600 pounds (90 gallons) were added on the day of the accident. 33 The left engine failure was allowed to operate as if the autolight system occasionally restored the engine briefly before failing again. The Safety Board pilots were briefed that the simulated engine failures would occur but did not know when. 31 NTSB/AAB-02/05 Statements from the carrier’s owner and the first officer’s fiancée and information from the weight and balance form completed by the accident crew suggest that the flight crew planned to add a total of 180 gallons to the airplane.

[ANALYSIS] Although low fuel quantity annunciator lights for each tank illuminate on the forward instrument panel when the fuel quantity decreases below 200 pounds, Safety Board investigators determined in simulations that these annunciator lights were easily overlooked, even when illuminated. Flight crews are required to constantly monitor their airplane’s fuel levels. If a pilot suspects that the fuel gauges are unreliable, fuel indicator sticks can be used to compare actual tank fuel levels to fuel gauge indications. Nevertheless, either pilot should have been able to determine a low fuel state by looking at the fuel gauges.35 Although the first officer’s fiancée stated that the first officer was concerned about the accuracy of the airplane’s fuel gauges, her statements and those of the operator’s CEO indicated that the tank levels had been measured before the airplane was fueled at FRG. It is possible that the pilots discounted the accuracy of the fuel gauges in flight because they had measured the tanks and believed that they added sufficient fuel (180 gallons) for the day’s flights. Further, it is also possible that the pilots 34 The “J31 Preliminary Checklist” contains a “Fuel Quantity” check item and a “low fuel caution system” test item. The “Before Takeoff Checklist,” from the “J31 Normal Checklist,” also has a fuel quantity check item. 35 The pilot who flew the accident airplane the day before stated that the fuel gauges were operating normally. 32 NTSB/AAB-02/05 trusted and monitored the LED display showing fuel used.36 However, the Safety Board notes that this display’s accuracy is dependent on pilot input of the correct fuel quantity at the beginning of the flight. Engine Stoppage Scenarios The accident airplane’s estimated fuel burn was calculated using information provided by the airplane manufacturer, including the Jetstream 31 flight manual, and the operator. Fuel burn calculations were based on flight times entered into the airplane’s logbook. Based on calculations for high-speed cruise power settings37 (which included fuel load and fuel burn time estimates for taxi, climb, en route flight, and descent), the accident airplane could have had about 275 pounds of fuel on board when it crashed. If the airplane, in fact, had 275 pounds of fuel evenly distributed between the tanks at the time of the crash this would not explain the right engine stoppage and subsequent crash. However, several factors could affect the accuracy of the assumptions that were used in these calculations and, consequently, the reliability/accuracy of the 275-pound result. Factors that can significantly change the amount of fuel burned on any flight include different (higher or lower) flight altitudes or airplane configurations, such as flaps and landing gear extended for long periods of time. Fuel burn is also increased during rapid, highperformance climbs, such as required for the execution of a missed approach. It is also possible that fuel load information for the accident airplane documented in the days before the accident did not accurately represent the airplane’s fuel load during this period. Carry-over errors in fuel load documents and differences in fuel calculations could result in assumptions of more fuel than was actually on board the airplane.

[ANALYSIS] The Safety Board pilots were briefed that the simulated engine failures would occur but did not know when. 31 NTSB/AAB-02/05 Statements from the carrier’s owner and the first officer’s fiancée and information from the weight and balance form completed by the accident crew suggest that the flight crew planned to add a total of 180 gallons to the airplane. If the flight crew intended to load 180 gallons (about 1,200 pounds), it was common industry and company practice to ask for 90 gallons on each side. However, based on the evidence, it appears that a lack of clear communication between the pilots and the fueler resulted in only 90 gallons (about 600 pounds) of fuel being added, a total amount confirmed by the fuel order receipt. Further, it is likely that the flight crew did not confirm the amount of fuel loaded before departure because flight crews do not see the fuel receipts after fueling and because the pilots were not outside the airplane monitoring the fuel loading. The accident flight crew completed a load manifest that stated that the airplane was loaded with 2,400 pounds of fuel when it departed from FRG, and it is probable that the accident flight crew planned to depart from FRG with 2,400 pounds. Based on the airplane’s performance, 2,400 pounds of fuel would have been sufficient fuel to fly the two flights without adding fuel during the stop at ACY. This would have allowed a takeoff from ACY that was within normal weight and balance limits for the airplane with 17 passengers on board. There was no evidence that the flight crew asked for additional fuel to be added during the stop at ACY. However, based on the Safety Board’s calculations, it appears that the accident airplane departed FRG with about 1,600 pounds of fuel on board, which is 800 pounds less than listed on the load manifest. Therefore, the Safety Board concludes that the accident airplane departed FRG with less fuel than the pilots thought they had on board. The accident pilots departed from FRG and ACY, completed two checklists (one for each of the flights) that included verification of fuel quantity,34 completed both load manifests for each of these flights (which required an entry indicating the amount of fuel on the airplane), and flew both flights apparently without being aware of a low fuel state. It is possible that the pilots’ belief that they had sufficient fuel on board for the flight to AVP caused them to be less vigilant in their monitoring of the airplane’s fuel state and slower to consider a low fuel state as the cause of the initial engine failure. Although low fuel quantity annunciator lights for each tank illuminate on the forward instrument panel when the fuel quantity decreases below 200 pounds, Safety Board investigators determined in simulations that these annunciator lights were easily overlooked, even when illuminated. Flight crews are required to constantly monitor their airplane’s fuel levels. If a pilot suspects that the fuel gauges are unreliable, fuel indicator sticks can be used to compare actual tank fuel levels to fuel gauge indications.

[ANALYSIS] The calculations for the bar differentials shown on the fuel quantity indicator are based on the 1,352-pound usable fuel capacity for each tank. As previously stated, beginning about 1331, the fuel supplied to the engine was being drawn solely from the right wing fuel tank by the right fuel boost pump. With the fuel return system operating normally, any excess fuel flow to the engine would have been returned and distributed equally to both fuel tanks.106 Thus, the NTSB concludes that the left and right fuel tanks were equally receiving fuel through the fuel return lines but that the left-wing-heavy fuel imbalance continued to increase during the flight because fuel was only being drawn from the right fuel tank. The increasing fuel level in the left tank and the accelerated depletion of the fuel from the right tank should have been apparent to the pilot because that information would have been presented on the fuel quantity indicator. The right fuel boost pump was providing all of the fuel system pressure to the engine because the left-side fuel system was no longer able to maintain the required fuel pressure through the operation of the left fuel boost pump. (At that time, the fuel pressure output was less than 2 psi, even with the left fuel boost pump on continuously.) The circumstances resulting in the low fuel pressure state and the degraded performance of the left-side fuel system (which restricted access to the fuel contained within the left wing tank and led to the left-wing-heavy fuel imbalance) were not resolved at any time during the flight; a possible cause for these conditions is discussed in section 2.2.4. Nevertheless, the NTSB concludes that the fuel system continued to provide fuel to the engine throughout the flight, even with the low fuel pressure state and the degraded performance of the left-side fuel system. 2.2.3.2 Descent Into the Terminal Area and Impact Radar data showed that, at 1402:52, the pilot changed the airplane’s route of flight and turned to the left toward BTM without ATC clearance. As shown in table 9, the pilot would have 106 About 1417, the left tank had reached its 1,368-pound total capacity (1,352 pounds of which was usable fuel), so any excess fuel from that point in the flight onward would have been vented overboard. The total amount of fuel vented overboard was estimated at 91 pounds. NTSB Aircraft Accident Report 56 seen a 15-bar differential on the fuel quantity indicator about that time; the maximum allowable fuel imbalance according to the PC-12 AFM was three bars. Also, CAWS data showed that, before the airplane’s course change, the pilot was unsuccessful in his attempts to laterally balance the fuel load by changing the status of the fuel pumps. The NTSB concludes that, although the pilot should have diverted to land the airplane as soon as practical after his attempt to restore fuel balance within the maximum imbalance limitation was unsuccessful, his eventual decision to divert to BTM likely resulted from his recognition of the magnitude of the increasing left-wing-heavy fuel imbalance situation.

[ANALYSIS] During the accident flight, the pilot did not begin to divert until about 20 to 31 minutes after his attempt to manually restore fuel balance was unsuccessful, and he chose BTM even though several closer airports along the airplane’s route of flight were available. The pilot had likely downplayed the seriousness of the initial advisories because no adverse outcomes resulted from ignoring the advisories during the flight from REI to VCB and during an October 2007 flight, when a low fuel pressure state existed that also necessitated the automatic operation of the left and right fuel boost pumps to provide the required fuel pressure to the engine (see section 1.16.2). Because the pilot did not take action to land the airplane as soon as practical after his attempt to manually restore fuel balance was unsuccessful, the fuel imbalance continued to worsen beyond the maximum imbalance limitation, increasing the airplane’s left rolling moment as the flight progressed. The accident sequence is further discussed in sections 2.2.2 through 2.2.6. 2.2.2 First Two Flight Legs The accident airplane had been fueled to its 406.8-gallon (2,736-pound) capacity on the day before the accident with 222 gallons (1,494 pounds) of Jet A fuel. The fuel truck at REI contained fuel that was not premixed with a FSII, but the fuel pump contained provisions for injecting a FSII during fueling. However, the pilot did not request that a FSII be added when the airplane was fueled. CORRECTED COPY NTSB Aircraft Accident Report 50 On the day of the accident, the pilot departed REI for VCB as the sole occupant of the flight. According to EIS trend data for the flight, the average outside air temperature was -32º C when the airplane was operating at its cruise altitude.93 About 1 hour 30 minutes into the flight, while the airplane was descending through an altitude of about 10,000 feet, the left and right fuel boost pumps began cycling (generally on for 10 seconds and off for 1 second). According to the Pilatus PC-12 AFM, the fuel boost pumps operate automatically if a low fuel pressure state exists—which occurs when fuel system pressure drops below 2 psi—and the pump’s switch is set to the AUTO position.94 By design, the CAWS annunciates a low fuel pressure caution when fuel system pressure drops below 2 psi for more than 0.3 second. However, because the CAWS did not log any fuel pressure cautions during the flight, it is likely that at least one of the fuel boost pumps was able to provide adequate pressure to the fuel system (at least 3.5 psi) within 0.3 second of the low fuel pressure condition being sensed. The fuel boost pumps continued cycling for 15 minutes, even though they were designed to turn off automatically 10 seconds after the fuel system pressure was restored to 3.5 psi.

[ANALYSIS] Also, about 3 minutes later (about 1418), the left fuel boost pump was on continuously, and the right fuel boost pump was cycling. In addition, at 2 hours 10 minutes into the flight (about 1420), the right fuel boost pump was on continuously, and the left fuel boost pump was off. Finally, about 6 minutes later (about 1426), the left fuel boost pump was on continuously, and the right fuel boost pump was cycling. 105 After the accident, Pilatus published an emergency procedure for low fuel quantity in the PC-12 AFM (see section 1.6.4.2). NTSB Aircraft Accident Report 55 Time in flight Local time Amount of fuel used (in pounds) Amount of total fuel remaining (in pounds) Imbalance (in pounds) Bar differential on fuel quantity indicator Left tank Right tank 1:13:32 1323 550 1,034 984 50 1 1:16:59 1326 570 1,033 965 68 1 1:17:59 1327 576 1,033 958 75 1 1:18:09 1328 583 1,030 955 75 2 1:21:05 1331 602 991 976 15 0 1:32:00 1342 667 1,087 814 273 5 1:52:08 1402 802 1,249 517 732 16 2:05:28 1415 881 1,367 320 1,047 22 2:07:00 1417 886 1,368 298 1,070 22 2:08:36 1418 898 1,368 274 1,094 23 2:10:34 1420 909 1,368 245 1,123 24 2:16:21 1426 944 1,368 160 1,208 25 2:17:05 1427 950 1,368 149 1,219 26 2:23:22 1433 979 1,368 66 1,302 27 Note: The calculations for the amount of fuel remaining and the fuel imbalance are based on the 1,368-pound total fuel capacity for each wing fuel tank. The calculations for the bar differentials shown on the fuel quantity indicator are based on the 1,352-pound usable fuel capacity for each tank. As previously stated, beginning about 1331, the fuel supplied to the engine was being drawn solely from the right wing fuel tank by the right fuel boost pump.

[PROBABLE CAUSE] Although the download of nonvolatile memory data provided key information in determining the circumstances that led to this accident, a flight recorder system that captured cockpit audio, images, and parametric data would have provided additional information about the accident that was not possible to determine from the downloaded nonvolatile memory data. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was (1) the pilot’s failure to ensure that a fuel system icing inhibitor was added to the fuel before the flights on the day of the accident; (2) his failure to take appropriate remedial actions after a low fuel pressure state (resulting from icing within the fuel system) and a lateral fuel imbalance developed, including diverting to a suitable airport before the fuel imbalance CORRECTED COPY NTSB Aircraft Accident Report 79

[ANALYSIS] The owner further indicated that the fuel sample that the pilot secured after the flight had ice crystals floating in it. Because the pilot subsequently serviced the airplane with fuel that contained a FSII and did not experience any further problems with the airplane, it is likely that the problem experienced during the flight resulted from the absence of a FSII, which caused the fuel system’s performance to be degraded. As previously stated, Pilatus required that a FSII be used for all flights operated with outside air temperatures below 0º C; the accident flight was operated in temperatures as low as -40º C. Evidence indicated that the accident pilot was aware of the need to add a FSII: the subject was discussed during recurrent training sessions, and fueling records between October 2007 and January 2009 indicated that the pilot always requested a FSII when fueling the airplane at San Bernardino.115 The NTSB could not determine why the pilot did not add a FSII during the four times that the airplane was fueled at REI in February and March 2009 and during the fueling at VCB on the day of the accident. The pilot was aware that at least two of the three flight legs on 114 Fuel records were not available to determine the amount of fuel that the airplane received at the FBO or whether a FSII had been added to the fuel. 115 The only other available fueling record obtained during this accident investigation showed a fueling that occurred in Riverside, California, on March 30, 2007. The airplane was fueled with 206 gallons of Jet A fuel; a FSII was not added to the fuel. However, it is not clear whether a FSII would have been required for the flights for that day. NTSB Aircraft Accident Report 62 the day of the accident would be conducted at high altitudes and in cold temperatures and should thus have understood the need for a FSII. In addition to the low fuel pressure state that existed during the accident flight, the left-side fuel system was not delivering fuel to the engine during much of the flight. The NTSB attempted to determine a possible reason for the restricted fuel supply from the left tank. Although the exact source of the restriction could not be identified, the postaccident testing clearly showed that ice accumulation in the fuel system (as a result of not adding a FSII) could degrade the performance of many fuel system components, including the fuel boost pumps and valves. The NTSB concludes that the low fuel pressure state and the restricted fuel supply from the left tank during the accident flight were the result of an accumulation of ice in the fuel system with an initial concentrated amount of ice at the airframe fuel filter. The NTSB further concludes that, if the pilot had added a FSII to the fuel for the flights on the day of the accident, as required, the ice accumulation in the fuel system would have been avoided, and a left-wing-heavy fuel imbalance would not have developed.

[CONCLUSIONS > FINDINGS] The airplane was controllable in static flight with the left-wing-heavy fuel imbalance that existed at the time of the accident, but the pilot lost control of the airplane with the dynamic maneuvers during the final moments of the flight. CORRECTED COPY NTSB Aircraft Accident Report 77 11. The large left rolling moment induced by the left-wing-heavy fuel imbalance could have been minimized or even avoided if the pilot had followed Pilatus Aircraft’s required procedures for flight operations with a fuel imbalance. 12. If the pilot had diverted earlier in the flight to one of several suitable airports along the airplane’s route of flight, the outcome of this flight would likely have been different because the airplane would have had a less severe fuel imbalance and the pilot would not have had to contend with the airplane’s deteriorating performance as the imbalance steadily progressed. 13. The pilot underestimated the seriousness of the initial fuel imbalance indications because he had not experienced any adverse outcomes from ignoring similar indications during previous flights. 14. The safety hazard involving fuel system ice accumulation could be mitigated if fuel filler placards installed aboard aircraft requiring a fuel system icing inhibitor specified that requirement. 15. Federal Aviation Administration pilot and operator guidance on the use of fuel system icing inhibitors would help raise awareness of the need to include this additive in turbine engine-powered aircraft fuel systems that require the additive. 16. At least four of the seven children on board the airplane were not restrained or were improperly restrained. 17. Although the number of passengers on board the airplane during the final flight leg did not comply with the PC-12 airplane flight manual limitation requiring no more than nine passengers, the four additional passengers on board the airplane did not directly affect the outcome of the accident. 18. For survivable accidents, passengers aboard airplanes operating under 14 Code of Federal Regulations Part 91 would be afforded better crash protection if each seat and restraint system were limited to only one passenger and children less than 2 years of age were restrained in an approved child restraint system. 19. Although the download of nonvolatile memory data provided key information in determining the circumstances that led to this accident, a flight recorder system that captured cockpit audio, images, and parametric data would have provided additional information about the accident that was not possible to determine from the downloaded nonvolatile memory data.

[ANALYSIS] Also, the right fuel boost pump was on continuously for 20 seconds and then resumed cycling. The brief change in the status of both fuel boost pumps was most likely induced by the pilot. Specifically, the pilot most likely recognized that the fuel imbalance was not being corrected and, as a result, repositioned the left fuel boost pump switch from ON to AUTO and the right fuel boost pump switch from AUTO to ON to observe the effect that the changed switch state would have on the fuel boost pumps’ 101 All fuel imbalance estimates in this analysis are based on the following assumptions: (1) an equal lateral fuel burn of 200 pounds during the flight from VCB to OVE, which would have resulted in 1,268 pounds of fuel in each tank when the accident flight leg began, and (2) a calculation of the fuel consumption and redistribution required for the activation of the R FUEL LOW caution toward the end of the flight (as discussed later in this section). 102 The Pilatus PC-12 AFM stated that fuel symmetry could be maintained (if the fuel balancing system were unable to maintain lateral fuel symmetry or if automatic fuel balancing were inhibited by a lateral fuel asymmetry that exceeded 270 pounds) by manually selecting the fuel boost pump switch to the ON position for the tank with the higher fuel quantity. The AFM indicated that, once a balanced fuel condition was restored, the fuel boost pump switch should be returned to the AUTO position. NTSB Aircraft Accident Report 54 operation.103 Similar switch repositioning activities occurred later in the flight, which were attributed to the pilot’s attempts to rebalance the lateral fuel load through manual activation of the fuel boost pumps.104 About 2 hours 17 minutes into the flight (about 1427), the CAWS annunciated a right fuel low (R FUEL LOW) caution; the Pilatus PC-12 AFM stated that this caution was logged when the amount of usable fuel in a tank (in this case, the right wing tank) was 133 pounds or less.105 The expected fuel load at this point in the flight, assuming 1,268 pounds of fuel in each tank (1,252 pounds of which was usable fuel) at the beginning of the flight and equal fuel consumption from each tank during the flight, would have been 812 pounds of usable fuel per tank. However, the fuel load at the time of the R FUEL LOW caution was estimated to be 1,368 pounds (1,352 pounds of which was usable fuel) in the left tank and 149 pounds (133 pounds of which was usable fuel) in the right tank, which corresponded to a left-wingheavy imbalance of 1,219 pounds. This imbalance would have been displayed to the pilot on the fuel quantity indicator as a 26-bar differential.

[ANALYSIS] The fuel boost pumps continued cycling for 15 minutes, even though they were designed to turn off automatically 10 seconds after the fuel system pressure was restored to 3.5 psi. Thus, it is likely that the fuel pressure dropped below 2 psi each time the boost pumps turned off, so the pumps continued cycling to return pressure to 3.5 psi within 0.3 second. After both fuel boost pumps had cycled for 15 minutes, the left fuel boost pump was on continuously, and the right fuel boost pump was off. This change in status occurred because, as the fuel boost pumps operated to correct the low fuel pressure situation, a left-wing-heavy fuel imbalance of about 70 pounds was created. According to Pilatus, when a fuel imbalance of about 70 pounds (5 percent of each wing’s total fuel capacity) occurs, the fuel boost pump in the tank with the higher fuel quantity operates automatically, and the fuel boost pump continues to operate until the fuel levels in both tanks are equalized. Pilatus also stated that a single fuel boost pump operating continuously could simultaneously correct a fuel imbalance and a low fuel pressure state. A fuel imbalance of about 70 pounds is displayed on the EIS fuel quantity indicator (see figure 4) by a two-bar differential between the fuel tanks. The left fuel boost pump was on continuously until the end of the flight, which occurred about 6 minutes later. The left fuel boost pump was able to maintain the required fuel system pressure during this time without the operation of the right fuel boost pump. Because the left fuel boost pump had not turned off before the end of the flight, it is likely that the fuel balancing system was still automatically rebalancing the fuel load at that time. When the airplane arrived at VCB, the pilot refueled the airplane at the airport’s self-service fueling island with about 128 gallons (861 pounds) of Jet A fuel.95 At this point, the fuel tanks would most likely have been refilled to their maximum capacity (1,368 pounds each). The fuel dispensed at the VCB self-service fueling island was not premixed with a FSII, and the 93 EIS trend data recorded between 0823:53 and 0905:05 PDT showed that, when the airplane was operating at FL220, the total air temperature was -24º C. The average outside air temperature, which is corrected for airspeed, was -32º C. 94 As stated in section 1.6.1, each fuel boost pump is controlled by a two-position (AUTO or ON) switch. The AUTO position is the normal setting, and the fuel boost pumps operate automatically when a pressure switch within the fuel system’s air separator detects fuel pressure below 2 psi.

[ANALYSIS] However, because the pilot had been employed with Eagle Cap Leasing for more than 6 years at the time of the accident, he should have been routinely using a FSII when servicing the airplane with fuel. On a standard day, the temperature is 0° C at 7,500 feet, so most of the flights conducted by the pilot for Eagle Cap Leasing (according to the airplane’s flight log) would have required the use of a FSII. Fueling records between 92 CAWS data showed that the accident airplane’s deice system was activated 2 hours 3 minutes into the flight (about 1413) and remained activated through the time of the last CAWS entry logged for the flight (about 1433). NTSB Aircraft Accident Report 49 October 2007 and January 2009 indicated that the pilot requested a FSII when fueling the airplane at its previous operating base, but fueling records from February and March 2009 showed that a FSII was not added during the four times that the airplane was fueled at its current operating base (REI). Analysis of data downloaded from the airplane’s CAWS showed that, during the flight from REI to VCB and the accident flight leg, a low fuel pressure state existed, which necessitated the automatic operation of the left and right fuel boost pumps to provide the required fuel pressure to the engine. The flight from REI to VCB concluded uneventfully. However, for the accident flight leg, the low fuel pressure state and a restricted flow of fuel from the left wing tank led to a left-wing-heavy fuel imbalance that exceeded 1,300 pounds, which ultimately resulted in a loss of control as the pilot maneuvered the airplane near the approach end of the runway. The airplane had been serviced with fuel on the day before the accident at REI and after landing at VCB, but the pilot failed to ensure that a FSII had been added to the fuel, even though he was aware that the flights would be operated at outside air temperatures that required the additive. Analysis of radar data showed that both flights were operating in below-freezing temperatures for almost their entire duration. In addition, Pilatus’ procedures required the pilot to (1) monitor the fuel quantity indicator in the cockpit to ensure fuel symmetry between the left and right fuel tanks during flight, and, in the event that fuel balance was not automatically maintained, (2) manually balance the fuel by turning ON the fuel pump for the fuel tank with the higher quantity, and, if unable to restore balance, (3) land the airplane as soon as practical. During the accident flight, the pilot did not begin to divert until about 20 to 31 minutes after his attempt to manually restore fuel balance was unsuccessful, and he chose BTM even though several closer airports along the airplane’s route of flight were available.

[ANALYSIS] Also, the fuel system configuration (which ensured a continuous, one-way flow of fuel to the engine, as shown in figure 3), along with the relatively high output pressure of the right-side fuel system, would have isolated any fuel flow provided by the left-side fuel system (because of the relatively low output pressure of the left-side fuel system at the time). As a result, the NTSB concludes that, about 1 hour 21 minutes into the flight, the fuel supplied to the airplane’s engine was being drawn solely from the right fuel tank by the right fuel boost pump, and the left-wing-heavy fuel imbalance continued to increase. The Pilatus PC-12 AFM stated that the fuel balancing system was designed to automatically correct fuel imbalances of up to 270 pounds, or about a six-bar differential on the fuel quantity indicator. If the left fuel boost pump had been automatically commanded by the fuel balancing system to turn on continuously, then the pump should have turned off once the fuel imbalance exceeded 270 pounds, which was estimated to have occurred about 1 hour 32 minutes into the flight (about 1342).101 The CAWS data recorded no discernible change in the left fuel boost pump’s status at this time (with the left fuel boost pump on continuously). Thus, the pilot had most likely recognized the fuel imbalance sometime between 1 hour 21 minutes and 1 hour 32 minutes into the flight and responded by repositioning the left fuel boost pump’s switch (located on the overhead panel) from its AUTO to ON position.102 This action allowed the left fuel boost pump to remain on continuously regardless of commands from the automatic fuel balancing system or the automatic activation of the pump in response to a low fuel pressure state. In addition, the Pilatus PC-12 AFM instructed pilots to monitor the fuel quantity indicator for fuel symmetry during each flight. About 1 hour 32 minutes into the flight, the fuel quantity indicator would have displayed a six-bar differential between the left and the right fuel tanks. The AFM stated that the maximum fuel imbalance for the PC-12 was 178 pounds with a maximum three-bar differential displayed on the fuel quantity indicator. The pilot elected to continue to BZN, even though the maximum three-bar differential had been exceeded. (This issue is further discussed in section 2.2.5.) The left fuel boost pump was on continuously and the right fuel boost pump was cycling until 1 hour 52 minutes into the flight (about 1402). At that time, the left fuel boost pump was off for 11 seconds and was then on continuously. Also, the right fuel boost pump was on continuously for 20 seconds and then resumed cycling. The brief change in the status of both fuel boost pumps was most likely induced by the pilot.

[ANALYSIS] No CAWS low fuel pressure cautions were logged during the flight, indicating that the low fuel pressure state was alleviated by the operation of one or both fuel boost pumps. About 1 hour 18 minutes into the flight (about 1328), the left fuel boost pump was on continuously, and the right fuel boost pump was off. The continuous operation of the left fuel boost pump indicated that the pump had been commanded to operate by the fuel balancing system to automatically correct a fuel imbalance of at least 70 pounds while the pump also operated to maintain adequate fuel system pressure.100 The fuel imbalance, which had developed between the left and the right fuel tanks (with the left tank containing more fuel than the right tank), would have been displayed by a two-bar differential on the fuel quantity indicator. This imbalance was likely created because the right fuel boost pump had delivered more fuel to the engine than the left fuel boost pump had delivered (likely as a result of a restricted flow of fuel from the left wing tank) during the time that both fuel boost pumps were simultaneously cycling. During the next 3 minutes, the operation of the left fuel boost pump had likely reduced the left-wing-heavy fuel imbalance to within about 15 pounds. About 1331, the right fuel boost pump resumed cycling. The operation of the right fuel boost pump indicated that (1) the fuel pressure output of the left-side fuel system had degraded to less than 2 psi, even with the left fuel boost pump on continuously; (2) the required fuel system pressure could no longer be maintained through the operation of the left fuel boost pump; and 98 The 3-minute 45-second duration of this activation was consistent with the average duration of 29 other automatic fuel boost pump activations that were downloaded from the accident airplane’s CAWS. The duration of the activation was also consistent with the 3-minute 42-second duration of the automatic fuel boost pump activation that was downloaded from the CAWS installed on N666M. All of these activations were considered normal lateral fuel balancing events because balancing 70 pounds of fuel within about 4 minutes is an expected behavior of the automatic fuel balancing system. 99 EIS trend data recorded between 1237:03 and 1405:24 showed that the airplane was operating at FL250 and with a total air temperature of -32º C. The average outside air temperature (corrected for airspeed) was -40º C. 100 CAWS data showed that the fuel balancing system detected the 70-pound imbalance about 1327 and that the left fuel boost pump activated about 1328 (because of the system’s 1-minute delay to determine whether an actual fuel imbalance was occurring). NTSB Aircraft Accident Report 53 (3) the right fuel boost pump was needed to maintain fuel system pressure to the engine.

[ANALYSIS] Some of these decisions ultimately affected the outcome of the accident flight. • On the day before the accident flight, the pilot had the airplane fueled at REI but did not request that a FSII be added to the fuel, even though he was aware that at least two of the flights on the day of the accident (REI to VCB and OVE to BZN) would include operations at high altitudes and in cold temperatures. The PC-12 AFM required a FSII for operations in outside air temperatures below 0º C. • Also on the day before the accident, the pilot filed three flight plans but misrepresented the number of occupants that would be on board the airplane during two of the flight legs.118 Specifically, the flight plans indicated that five occupants, including the pilot, would be on board for the flight from VCB to OVE (10 occupants would actually be on board) and that nine occupants, including the pilot, would be on board for the planned flight from OVE to BZN (14 occupants would actually be on board). The pilot was aware that the airplane would have nine seats available to the 13 passengers and should have been aware of the AFM limitation allowing no more than nine passengers to be transported aboard a PC-12. • For two of the three flight legs on the day of the accident, the pilot allowed the airplane to depart with a takeoff weight that was over the PC-12 maximum takeoff weight (see table 2). • After arriving at VCB on the day of the accident, the pilot used the self-service fueling station to add fuel to the airplane, but he did not add a FSII to the fuel. • The maximum allowable fuel imbalance between the left and the right fuel tanks was estimated to have been exceeded sometime between 1331 and 1335, and the pilot attempted to manually correct the imbalance sometime between 1331 and 1342. The PC-12 AFM stated that, if the fuel cannot be balanced, the pilot should land the airplane as soon as practical. The pilot did not divert to another airport at that time, even though three suitable airports along the airplane’s route of flight—BOI, TWF, and LLJ—were available to the pilot. • The pilot began to divert to BTM about 20 to 31 minutes after his attempt to restore fuel balance was unsuccessful. At that time, LLJ was the closest airport to the airplane’s position. Once the airplane’s route of flight changed, DLN became the most suitable diversion airport relative to the airplane’s position, but the pilot decided to continue to BTM. The NTSB evaluated fatigue as a possible factor in the pilot’s decision-making process. Limited information was available about the pilot’s sleep history in the days before the accident.

[CONCLUSIONS > FINDINGS] However, the captain’s calculations were incorrect, and the airplane’s weight and balance were within limits. 8. The captain did not recognize the importance of the cues provided by the first officer, and he failed to perform expected pilot-in-command duties. 9. The captain was preoccupied with the weight and balance calculations during critical portions of the flight and, as a result, he did not monitor the fuel crossfeed operations, which resulted in a fuel imbalance and unusual airplane handling characteristics. 10. The flight crew did not monitor the fuel quantity gauges or respond properly to the airplane’s changing handling characteristics, and the captain did not start the in-range checklist at the appropriate altitude; as a result, the crew missed several opportunities to identify the ongoing fuel crossfeed operations and determine that the airplane’s fuel was imbalanced. Conclusions 38 Aircraft Accident Report 11. Although fuel transfer is prohibited on the Convair 580 airplane, fuel transfer can occur during fuel crossfeed operations if the fuel tank shutoff valve for the tank not being used is left open. 12. All of the fuel from the airplane’s left tank that was not used by the engines transferred into the right tank because the captain intentionally kept the right fuel tank shutoff valve open during fuel crossfeed operations, which was not in accordance with approved fuel crossfeed procedures. 13. During the airplane’s descent to landing, the fuel in the left fuel tank, which was providing fuel to both engines, was exhausted because both engine-driven fuel pumps drew air from the left tank into the fuel system instead of fuel from the right tank, resulting in a dual-engine flameout. 14. Fuel transfer can occur on the Convair 580 airplane if it is operated with different fuel boost pump output pressure settings and with the fuel crossfeed valves unintentionally left open. 15. The accident investigation would have benefited from the retrofit of an independent cockpit voice recorder power source and a cockpit image recorder. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was fuel starvation resulting from the captain’s decision not to follow approved fuel crossfeed procedures. Contributing to the accident were the captain's inadequate preflight planning, his subsequent distraction during the flight, and his late initiation of the in-range checklist. Further contributing to the accident was the flight crew’s failure to monitor the fuel gauges and to recognize that the airplane’s changing handling characteristics were caused by a fuel imbalance. 39 Aircraft Accident Report

[ANALYSIS] However, manufacturer and FAA SDR data did not indicate a history of Convair fuel tank shutoff valve failures. 57 At the time of the accident, Nolinor Aviation’s FOM stated that the fuel tank valve on the tank to be shut off should be closed, “subject to the captain’s discretion.” After the accident, Nolinor removed this remark from the crossfeed procedures contained in its FOM. Analysis 33 Aircraft Accident Report concludes that all of the fuel from the airplane’s left tank that was not used by the engines transferred into the right tank because the captain intentionally kept the right fuel tank shutoff valve open during fuel crossfeed operations, which was not in accordance with approved fuel crossfeed procedures. Therefore, to prevent similar accidents from occurring in the future, the Safety Board believes that the FAA should issue a flight standards information bulletin to all principal operations inspectors of Convair 580 operators that familiarizes operators with the circumstances of the Air Tahoma flight 185 accident, including the importance of closing the fuel tank shutoff valve for the tank not being used during fuel crossfeed operations. 2.5 Dual Engine Power Loss The Safety Board considered why both engines lost power even though the right fuel tank had sufficient fuel for operation. Further, after the engines failed, the flight crew was still able to make transmissions to ATC, the GPWS alerted, and the FDR continued to record, indicating that the airplane still had partial electrical power. Wreckage examinations indicated that, when the left fuel tank was exhausted of fuel, the left fuel boost pump most likely continued to operate, which allowed air to enter the fuel system and reduced the fuel pressure to the left and right engine-driven fuel pump inlets. If the captain had turned on the right fuel boost pump after he identified that he had left the crossfeed valve switch open, it is possible that the right engine would have received sufficient fuel pressure to continue to operate; however, there is no evidence that he did so. Specifically, the right fuel boost pump fan motor cooling fan cover and the blade directly beneath the cover were found bent inward, and no rotational markings were found, indicating that the right fuel boost pump was not operating at the time of impact. Postaccident testing of the left and right fuel pumps revealed no evidence of preexisting failures.

[ANALYSIS] The Safety Board concludes that the captain did not recognize the importance of the cues provided by the first officer, and he failed to perform expected PIC duties. At 0045:37, shortly after the airplane passed through 3,200 feet, the captain started the in-range checklist, which includes a step to check the fuel tank shutoff and crossfeed valve switch positions. Air Tahoma’s FOM indicates that the in-range checklist should be started before leaving 12,000 feet and that “early completion [of the checklist] will minimize cockpit distractions.” Therefore, the captain should have started the in-range checklist about 14 minutes earlier (shortly before 0035; at which time, the airplane was at an altitude of about 12,000 feet, and the first officer had not yet stated his concerns about the airplane’s unusual handling characteristics). While performing the in-range checklist, the captain stated that they had an “imbalance on this…crossfeed I left open.” This was the only comment recorded by the CVR regarding fuel crossfeed operations since the captain started the operation almost 30 minutes earlier. Further, the CVR did not record any comments by either flight crewmember indicating that they were monitoring the airplane’s fuel quantity. If the flight crewmembers had been monitoring the gauges, they would have seen unmistakable indications of a developing fuel imbalance. The fuel imbalance developed over a 30-minute period, and, as noted, during that time, the pilots had several opportunities to recognize and correct the problem; however, they failed to do so. The Safety Board concludes that the captain was preoccupied with the weight and balance calculations during critical portions of the flight and, as a result, he did not monitor the fuel crossfeed operations, which resulted in a fuel imbalance and unusual airplane handling characteristics. The Safety Board further concludes that the flight crew did not monitor the fuel quantity gauges or respond properly to the airplane’s changing handling characteristics, and the captain did not start the in-range checklist at the appropriate altitude; as a result, the crew missed several opportunities to identify the ongoing fuel crossfeed operations and determine that the airplane’s fuel was imbalanced. At 0046:35, the first officer stated, “we’re gonna flame out.” The captain responded, “I got the crossfeed open.” The first officer then stated, “we’re losing power,” and, “we’ve lost both of them [engines].” The captain replied, “nope.” The CVR stopped Analysis 32 Aircraft Accident Report recording at 0046:55, and, about the same time, a momentary interruption in electrical power occurred. About 2 minutes later, the FDR stopped recording. Although Air Tahoma’s QRH contained engine flameout procedures, the pilots most likely would not have had sufficient time or altitude to restart the engines because they lost power during short final approach. 2.4 Convair 580 Fuel Crossfeed Operations The Convair 580 type certificate data sheet and Air Tahoma’s FOM prohibit fuel transfer from one tank to the other while the airplane is on the ground or in flight.

[ANALYSIS] The envelope in which the lease agreement was mailed had affixed to it Pitney Bowes Meter postage, dated October 17, 1977, at Addison, Texas. The Pitney Bowes postage was cancelled by the Dallas, Texas, Post Office during "pm" of October 21, 1977. 1.18 New Investigative Techniques None 2. ANALYSIS The flightcrew was properly certificated and trained in accordance with applicable regulations. There was no evidence of preexisting medical problems that might have affected the flightcrew's performance. The aircraft was certificated and equipped according to applicable regulations. The gross weight and c.g. were within prescribed limits. The aircraft's structure and components were not factors in this accident. There was no evidence of any malfunction of the aircraft or its control system. The propulsion system was operating and was producing power until fuel was exhausted. The right engine had been malfunctioning for some time and caused the flightcrew to operate that engine on auto-rich fuel mixture during the accident flight and. during previous flights in order to obtain an acceptable level of performance from it. Although examination of the engine and its components did not identify the exact discrepancy, the Safety Board believes that the discrepancy was of a general nature, such as an ignition or induction problem, and was not a major mechanical failure. Components of the right engine's ignition system were so badly damaged by impact that engine to distributor timing could not be determined. Consequently, the pre-impact condition of the ignition system could not be determined from the evidence available. Based on wreckage examination, the Safety Board concludes that both engines ceased producing power because of fuel exhaustion. Only one quart of fuel was recovered from both engines. Evidence obtained from the fuel quantity gages indicates that both fuel tanks were empty at the time of impact. According to the best estimates, the aircraft should have had about 207 gallons of fuel on board at the time of the accident. This figure is based on a normal cruise configuration with both engines operating with "auto-lean" fuel mixture. In order to determine the reason for the discrepancy between calculated fuel on board and actual fuel on board, the Safety Board analyzed the following three explanations: First, there could have been a fuel leak. However, no evidence of fuel leakage, such as stains or loose fuel tank caps or lines, was found in the wreckage. Although this possibility cannot be discounted completely, because there is a remote possibility that leakage evidence could have been obliterated at impact, the Safety Board does not believe it to be the most viable explanation. Secondly, the aircraft may not have been fueled with the amount shown on the fuel slips. The Safety Board considers this explanation relatively remote because the fuel meters on refueling trucks cannot be reset and, if functioning properly, will reflect the total amount of fuel dispensed to a given aircraft. Finally, the engines or an engine could have been burning more fuel than specified and more than the flightcrew expected to be burned.

[ANALYSIS] Secondly, the aircraft may not have been fueled with the amount shown on the fuel slips. The Safety Board considers this explanation relatively remote because the fuel meters on refueling trucks cannot be reset and, if functioning properly, will reflect the total amount of fuel dispensed to a given aircraft. Finally, the engines or an engine could have been burning more fuel than specified and more than the flightcrew expected to be burned. The witnesses report of torching from the right engine would indicate a rich fuel mixture or other discrepancy associated with inadequate combustion. Operating an engine in the auto-rich configuration would increase the fuel consumption by about 25 gallons per hour for that engine, from 183 gal/hr to 208 gal/hr. During the accident flight of 2.8 hours this would have amounted to about 70 gallons. It is impossible to determine how long the aircraft was operated with the right engine in auto-rich, but it was evidently long enough to exhaust the useable fuel on board the aircraft. Regardless of the high fuel consumption of the right engine, the 5-hour, or 900+ gallon, fuel supply listed on the flight pian would have been sufficient to reach the destination. Considering the increased fuel consumption on the right engine, 583 gallons would have been required to complete 2.8 hours of flight from Greenville to the accident site. Therefore, the Safety Board concludes that the right engine was burning more fuel than anticipated because it was being operated in the auto-rich fuel mixture. The crew was either negligent or ignorant of the increased fuel consumption because they failed to monitor adequately the engine instruments for fuel flow and fuel quantity. Had they properly monitored their fuel supply and noted excessive fuel consumption early in the flight, they could have planned an alternate refueling stop rather than attempting to continue flight with minimum fuel. In addition, the Safety Board believes that the pilot was not prudent when he continued the flight with a known engine discrepancy and did not have it corrected before he left Greenville. This accident involves another operation where the party which had operational responsibility is in controversy. It appears that it was the intent of L & J Company to have the operational responsibility assumed by the lessee, Lynyrd Skynyrd Productions. The lessee, however, appears to have had no understanding that it was the operator and had assumed the responsibilities thereby imposed. The question of who was the legal operator of this flight is currently being litigated by the FAA in an enforcement action against L & J Company and will be addressed in civil litigation arising out of this accident. In examining this relationship, the Board reviewed the lease for this flight.

[FINDINGS] If this occurs, it should be a step toward resolving the problem of the uninformed lessee. 3. CONCLUSIONS’ 3.1 Findings l. Both engines of N55VM ceased to produce power because the aircraft's useable fuel supply was exhausted. 2. The crew failed to monitor adequately the fuel flow, en route fuel consumption, and fuel quantity gages. 3. The crew failed to take appropriate preflight and maintenance action to assure an adequate fuel supply for the flight. 4, The crew operated the aircraft for an indeterminate amount of time before the accident with the right engine's mixture control in the auto-rich position. -~ 16 - 5. There were no discernible discrepancies between the amounts of fuel added to the aircraft and the amounts shown on the fuel receipts from the servicing facilities. 6. There was no evidence of a fuel leak. 7. There was no fire after impact because little fuel remained in the aircraft's fuel system. 8. The survival of many passengers was due to the lack of severe impact deformation in the center of the fuselage and the absence of a postcrash fire. 9. The provisions of the lease intended to satisfy the requirement for a "truth in leasing clause" did not result in this lessee having an adequate understanding as to who was the operator of this flight and what that means. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was fuel exhaustion and total loss of power on both engines due to crew inattention to fuel supply. Contributing to the fuel exhaustion were inadequate flight planning and an engine malfunction of undetermined nature in the right engine which resulted in higherthan-normal fuel consumption. 4. SAFETY RECOMMENDATIONS No safety recommendations were submitted as a result of this accident. FAA issued Advisory Circular 91-37A on January 16, 1978, with detailed guidance relative to leasing of aircraft. BY THE NATIONAL TRANSPORTATION SAFETY BOARD /s/ JAMES B. KING Chairman /s/ FRANCIS H. McADAMS Member /s/ PHILIP A. HOGUE I Member /s/ ELWOOD T. DRIVER Member June 19, 1978 ~i7- 5. APPENDIXES APPENDIX A INVESTIGATION Investigation At 2025 e.s.t. on October 20, 1977, the National Transportation Safety Board was notified of the accident by the FAA Communications Center in Washington, D.C. An investigative team was dispatched immediately to McComb, Mississippi, and working groups were established for operations, human factors, structures, systems, powerplants, air traffic control, witnesses, weather, and aircraft records. The Federal Aviation Administration, Convair Division of General Dynamics, and Pratt & Whitney Aircraft Group of United Technologies participated in the investigation.

[PROBABLE CAUSE] Evidence indicates that the fuel quantity indicating-system accurately indicated fuel quantity to the crew. The fuel gages are readily visible to the captain and the second officer. The captain failed to make decisive timely decisions; The captain failed to relate time, distance from the airport, and,the aircraft’s fuel. state. as hisattentiqn-was directed cbmple tely to.ward the ,_.. _..- _ . ..- + -- . ’ --- , - 3.2 Probable Cause The National Transportation Safety Board determined that the probable cause of the accident was the failure of the captain to monitor properly the aircraft’s fuel state and to properly respond to the low fuel state and the crew-member’s advisories regarding fuel state. This resulted in fuel exhaustion to all engine‘s. His inattention resulted from preoccupation with a landing gear malfunction and preparations for a possible landing emergency. Contributing to the accident was the failure of the other two flight crewmembers either to fully comprehend the criticality of the fuel state or to successfully communicate their concern to the captain. I ;j *30- - ’ * _.__ , , 4 4. Safety Recommendations As a result of this accident, the Safety Board has issued the following recommendations to the Federal Aviation Administration: “Issue an .Operations Alert Bulletin to have FAA inspectors assure that crew training stresses differences in fuel-quantity measuring instruments and that crews flying with the new system are made aware of the possibility of misinterpretation of gage readings. (Class II--Priority Action) (A-79-32)” “Emphasize to engineering personnel who approve aircraft engineering changes or issuance of Supplemental Type Certificates the need to consider cockpit configuration and ” instrumentation factors which can contribute to pilot confusion, such as the use of similar-appearing instruments with different scale factors. (Class II--Priority Action) (A-79-33)” “Xudit Supplemental Type Certificate SX3357 \\‘E-D for completeness, e-pecially in the area of system calibration after installation. (Class II--Priority .4ction) (X-79-33)” “Issue an operations bulletin to all air carrier operations in-spec tors directing them to u~c their assigned operators to ensure that their flightcrews are indoctrinated in principles o f f l i g h t d e c k resoume management, with particuhr emphasis on the merits of participative management for captains and assertivene.ss training for other cockpit crewmembers. (Class II, Priority diction) (X-79-17)” BY THE NXTlONXL TR.\NSPORTATlON SAFETY BOARD /S/ JAMES B. KINGChair man /s/ ELWOOD T. DRIVER Vice Chair man .,:. _’ I June 7, 1979 /s/ FRANCIS H. McADAMS Member /s/ PHILIP A. --- Footnotes: [S/ N 669234 No, 2 Engine 10-24-78 27 .685:28 11,897 597:43] [s/ Aeronautical Radio, Inc., an air-to-ground radio service which provides a communication system for commercial aircraft.]

[ANALYSIS] Proper crew management includes i constant awareness of fuel remaining as it relates to time. In fat t, the Safety Board believes that proper planning would provide for enough fuel on landing for a ,) go-around should it become necessary. Such planning should also consider possible ’ fuel-quantity indication inaccuracies. < This would necessitate establishing a deadline time for initiating the approach and constant monitoring of time, as well as 4: the aircraft’s position relative to the active runway. Such procedures I a’ _-- ,* + -- al should be routine for all flightcrews. However, based on available evidence, this flightcrew did not adhere to such procedures. On the contrary, the ssx9q-k period of ku9ding. The other two flight crewmembers, although they made several hg time to fuel exhaustion would have been voiced. However, there was none until after the aircraft was already in a position from which recovery was not possib$ In analyzing the flightcrew’s actions, the Safety Board considered that the crew could have been misled by inaccuracies within the fuel-quantity meatiiing system. However, those intracockpit comments and radio transmissions in which fuel quantity was’mentioned indicate that the fuel-quantity indicating system wa.s a c c u r a t e . A A Had the flightcrew reld&my 91. thc>$--&el qk_an_tit&s-.to fuel flow, -they should have been aware_that fuel exhaustion wouldxcur at oa~~t.l815f Other p-” - - evidence that the captain had failed to assess the effect of continued holding on fuel state was provided by his stated intentions to land about 1805 with 4,000 Ibs of fuel on board. Just minutes earlier, at 1748:56, he was made aware that only 5,000 lhs remained. During the 16 min between the observation of 5,000 lbs and 1805, the aircraft would consume at least 3,000 Ibs of fuel. Further evidence of .the flightcrew’s lack of concern or aivareness was provided when observations of 4,000 lbs remaining about 17 min before the crash, the left the cockpit at the captain’s request to check on the cabin evacuation preparations. Upon his return, about 4 min later, he gave the captain an estimate of another 2 or 3 min for the completion of the cabin preparation. At this time;‘.the aircraft was in the general vicinity of the airport. In the initial interview with the captain, he stated that he felt the cabin preparatmd be ea m rrom IO-0 mm and fiat the “taiIXiaafif’~could‘be accomplish;;d roach to the airport.

[ANALYSIS] According to the tape of the conversation between the captain, the company dispatcher, and company line maintenance personnel, the captain had advised the dispatcher that he had 7,000 lbs of fuel aboard and that he intended to land in 15 or 20 min. The dispatcher then checked with the captain to ascertain a specific time for the landing and the captain agreed that 1805 was Ita good ballpark.” The dispatcher, according to his interview after the accident, then relayed this landing time and the aircraft’s status to the company personnel in Portland. He also stated that his assessment of the situation was that of the fuel remaining upon landing would be low but the landing could be made successfully at 1805. The Safety Board believes that, with the information given to him by the captain, the dispatcher acted properly and in accordance with company procedures. P 3. CONCLUSIONS 3 .l Findings 1. 2. 3. 4. 5. 6. 7. a. 9. 10. 11. 12. 13. 14. Except for the failure of the piston rod on the right main landing gear retract cylinder assembly, with the resulting damage to the landing gear position indicating system switch, there was no evidence of a failure or malfunction of the aircraft’s structure, powerplants, flight controls, or systems. : The aircraft departed Denver with the required fuel aboard of 2 hrs 26 min for the en route flight and with the required E:.4R and company contingency fuel aboard of about 1 hr. The aircraft began holding about 1712 at 5,000 ft N ith its gear doivn; this was about 2 hrs 24 min after it departed Denver. The flightcrew was properly certificated and qualified for the flight. The aircraft was certificated, maintained, and dispatched in accordance with Federal Aviation Regulations and approved company procedures. The landing delay covered a period of about 1 hr 2 min. All of the aircraft’s engines flamed out because of fuel exhaustion about 1815-l hr 3 min after it entered into hold and 3 hrs 27 min after it departed Denver. Fuel exhaustion was predictable. The crew failed to equate the fuel remaining with time and distance from the airport. No pertinent malfunctions were found during examinations of the fuelquantity measuring system. A new digital fuel-quantity indicating system was installed on this aircraft on hlay 12, 1978. This was in accordance with a DC-8 UAL flee tw ide retrofit program. Evidence indicates that the fuel quantity indicating-system accurately indicated fuel quantity to the crew. The fuel gages are readily visible to the captain and the second officer. The captain failed to make decisive timely decisions; The captain failed to relate time, distance from the airport, and,the aircraft’s fuel. state. as hisattentiqn-was directed cbmple tely to.ward the ,_.. _..- _ . ..- + -- . ’ --- , -

[(a) FINDINGS] The aircraft fuel cap cover plates were ' Likewise marked legibly to indicate the type of fuel required. Both engines failed to produce adequate power during climb because of overheat and severe detonation, ‘It is apparent that the linemen had not received sufficient training or experience to recognize the fuel requirements of the aircraft. The fact that they added 57 quarts of reciprocat~ ing engine of' should have alerted them to their error. The first officer who was monitoring the fueling operation failed to observe the improper fueling of the aircraft, The first officer ‘cho personally drained the cel tank sumps, failed to detect the presence of an oily substance as did the fueler. Frobable Cause The Safety Board determines that the probable cause of this accident was the loss of effective engine power because of improper fel having been placed in the tanks by relatively. untrained persomiel. A contributing factor was that the flightcrew did not detect the error. 3. RECOMMENDATIONS On the basis of this investigation, the Board has vecommended to the Administrator, Federal Aviation Administration, that Parts 23, 25, 27 and 29 of the Federal Aviation Regulations and Advisory Circular 20-43a be amended to provide a more adequate color coding system for aircraft yvefueling. (See appendix D.) BY THE NATIONAL TRANSPORTATION SAFETY BOARD: /sf JOHN H, REED Chairman — /s/ OSCAR M. LAUREL Member /s/ FRANCIS H. McADAMS Member /s/ LOUIS M. TKAYER Meraber Isabel A. Burgess, Meaber, did not participate in the adoption of this report, : September 30, 1970. APPENDIX 4 INVESTIGATION AND HEARING 1. Investigation The Board received notification of the accident at approximately 1045 e.d.t., May 3U, 1970, from the Federal Aviation Administration, An investigating team was dispatched immediately to the scene of the accident. Working groups were established to conduct the factfinding processes in the areas of: Operations, Air Traffic Control, Weather, Structures, Systems, Powerplants, Witnesses, znd Human Factors. Participants in the investigation were representatives of the Federal Aviation Administration, Pratt & Whitney Aircraft, Airwork Service Division. The on-scene phase of the investigation lasted approximately 5 days. 2. Public Hearing A public hearing was not held in connection with the investigation of this accident. 3. Preliminary Reports There were no preliminary reports issued in connection with this accident. APPZNDIX. B Crew Information Captain James A. Cannin, aged 57, held Airline Transport Pilot certificate No. 57806. He also held ratings in Douglas DC-3, DC-4, DC-6~7, Martin 202/404 and Curtiss C-46.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] One hundred gallons of Jet A fuel were serviced into each of the aircraft's two fuel tanks. The introduction of the jet fuel resulted in a fuel mixture that was not compatible with the engines installed. The introduction of the jet fuel lowered the octane rating of the fuel resulting in high opera: ing ‘temperatures, severe detonation, and extensive and sustained power loss. While engine operation was adequate to accomplish a successful takeoff with the utilization oi the water injection system, the termination of water supply to the engine subsequent to takeoff resulted in immediate manifestation of the detonation and/or preignition conditions. The apparent inability of the flightcrew to recognize immediately the symptoms of detonation, or to assess the cause for this condition, prompted the application of carburetor heat which further elevated induction temperatures and contributed to even more adverse detonation condition. Cylinder head tenperatures of both engines attained the 300° C. gage limit as a result of the detonation and/or preignition, Again the action taken in opening the cowl flaps, in an effort to rectify this indication, was responsible for increasing drag which adversely affected the critical thrust/drag configuration of the aircraft, There was no clear understanding between the crew and the fuel service personnel regarding the type of fuel to be delivered. I It is apparent that the linemen had not received sufficient training or experience to recognize the fuel requirements of the aircraft. Both linemen knew they were servicing the aircraft with jet fuel; therefore, they evidently did not identify the type aircraft or engines installed, I The mere fact that they added 57 quarts of reciprocating engine oit should have alerted them to their error. ‘The first officer, had he been observant, could have seen the fuel servicing unit markings. ‘The price of the fuel delivered and the type fuel "JET" on the sales - invoice should have alerted the captain of the servicing error, 2.2 Conclusions (a) Findings ', The flightcrew members were certificated properly and qualified for the operation involved. 2. The aircraft was airworthy both mechanically and structurally. Its gross weight and center of gravity were within Limits at time of departure. . There was no indication of wechanical failure or malfunction of the aircraft stcucture or systems, The aircraft was operated in Instrument Meteorological Conditions (IMC), while on an IFR clearance. There were no difficulties with navigational aids, communications, or ground based radar equipment. ‘There was no clear understanding between the crew and fuel service personnel regarding the type of fuel to be delivered. The aircraft was serviced with 200: gallons of Jet A fuel which diluted and lowered the octane rating of the fuel on board to a point that it was not compatible with the engines installed, The fuel service unit was marked conspicuously to indicate that ft contained jet fuel. The aircraft fuel cap cover plates were ' Likewise marked legibly to indicate the type of fuel required. Both engines failed to produce adequate power during climb because of overheat and severe detonation, ‘It is apparent that the linemen had not received sufficient training or experience to recognize the fuel requirements of the aircraft.

[ANALYSIS] When questioned, the manufacturer's laboratory conceded that this test should not be considered conclusive, and that standard test methods should be relied upon. In addition, the airplane was fueled at Anchorage, as well as at previous refueling stations in the Reeve system, with Jet A fuel. Therefore, it would have been impossible to have JP-4 fuel in the fuel controls, Therefore, the Safety Board concludes that the fuel was Jet A. The Safety Board further concludes that the water in the fuel was net the result of any improper handling, but rather was the result of dissolved water coming out of solution in a suspended state due to the unusually low fuel temperatures. -1]?- 2.3 Loss of No. 2 (Right) Engine Power The loss of power on the No. 2 (right) engine, indicated by the low torque and fuel flow, can be associated directly with the reinoval of fuel heat as part of the before-landing check. The first officer first noticed the loss of torque after he turned off the fuel heaters. Because of the lengthy exposure of che airplane to very low temperatures before flight and to extremely low temperatures during cruise, the fuel temperature would have been well below the freezing point of water. The draining of the fue) sumps before and after refueling did not reveal any free water in the tanks. However, this visual check, while necessary to insure the absence of free water, cannot reveal water dissolved in the fuel and must not be relied upon as indicative that fuel icing will not occur, Analysis of the fuel samples, taken after the accident in less than ideal conditions, confirmed the presence of water in the fuel system. It is likely that the actual quantity of water, both free and dissolved, was greater than that implied by the fuel analysis, Because the samples were taken when the ambient temperatures were still well below 0° F, anv collected free water in the system would have been frozen and would not have flowed through the drains when the samples were taken. During flight, the water would hav< been dispersed in the fuel either as supercooled droplets or small ice crystals in the fuel, but would not have restricted the fuel flow in the system in that form. And so long as heat was supplied by the fuel heater, the water would have easily passed through the engine fuel system as liquid. However, the fuel in the airplane's tanks was probably at very tow temperatures because of the long exposure to low ambient temperatures. Therefore, the fuel heater would have raised the temperature of the fuel entering the engine only slightly above freezing. Once the fuel heaters were turned off, the water entrained in the cold fuel would have frozen quickly in the small orifices and screens in the fuel control and pump, and also could have impregnated the filter element and partially blocked fuel flow through the filter. Although the first officer, when interviewed, seemed to associate the loss of torque with the removal of fuel heat, it seems apparent from the CVR transcript that he did not do so at the time, and, therefore, did not attempt to restore fuel heat. The captain stated that he sensed the loss of torque at the same time the first officer noticed the low torque indication.

[ANALYSIS] It has been shown by previous tests by the manufacturer that water in the . fuel that freezes in the fuel contro! servo system ean cause large fuel flow fluctuations. If the orifices or screens in the servo system were blocked, moving the throttle would cause fuel pump output flow to increase, but the servo signal would be prevented from balancing the fuel pump output with reference to throttle position. Thus, the fuel Mow to REE oe eet ; the engine would become excessive and could produce an overtemperature condition. ke Because the actual positions of the high-pressure cock levers and the cruise pitch stops * could not be verified by the available evidence, and because the presence of ice in the :: fuel control could not be verified but only inferred from the water found in the control, E. the Safety Board is unable to determine the precise cause of the overtemperature. - Although the Safety Board could not establish the exact cause of the left 4 engine failure, it is apparent that this accident sequence originated with the removal of o fuel heat in accordance with the operating manual and defore-landing checklist. The ee Safety Board believes that while the crew complied with the manual instructions and i checklist, they did so without a full understanding of the significance of fuel temperatures that were well below freezing. The instructions in the Nihon operations manual and the Reeve training manual are not sufficiently detailed to provide adequate guidance for the crew to make a proper decision on the need for continuing the use of fuel heat. As this accident revealed, the requirement in the manual for using fuel heat for 2 minutes within 5 rninutes of landing is not adequate to prevent engine fuel system icing when operating in extreme conditions. The Safety Bourd believes that if the manuals and crew training hed Oa EL eemiemeaiameinamemnateaiabeatediainthnriotseameeahatianeres hte teaateanem inated es ee ie ee ~14- included mere specific discussion and instructions for the use of fuel heat, the crew would have been aware of the need to make a decision on whether fuel heat was still required during the approaen and also would have had the guidarce necessaty to make that decision. Ee OT: ces me ne me eae Rie eta nett 2.5 Emergency Evacuation Passengers were not able io open the right overwing exit to escape from the airplane until assisted by the captain who opened the exit. Subsequent investigation revealed that the catch wes difficult to locate. The Safety Board believes that under conditions of panic, poor lighting, or fire with its accompanying smoke and toxie fumes, where time for escape is severely limited, a person unfamiliar with the exit operation would have difficulty locating the catch and subsequently opening the exit. Oe ne he he a The type certificate for the YS-11 was applied for on June 15, 1962, and was approved on September 7, 1965.

[ANALYSIS] The first EFC was for 2030, issued about 8 minutes before the flight entered holding at CAMRN, and the second EFC was for 2039 as the flight entered holding at CAMRN. The flightcrew may have assumed that the previous EFC’s were valid times for which they would receive clearance to depart holding and begin the approach to JFK before the fuel state became more critical. In fact, EFC’s are merely estimates by the controllers based on a dynamic traffic and weather situation and are issued to provide a time to commence the approach should the flight lose radio contact. When ATC issued a third EFC of 2105, the flightcrew apparently finally realized that they had to commence an approach and therefore requested priority handling. However, the Safety Board concludes that the flightcrew had already exhausted its reserve fuel to reach its alternate by the time it asked for priority handling. When asked a second time for its alternate, the first officer responded, at 2046:24, "It was Boston, but we can’t do it now, we, we, don’t, we run out of fuel now." Although the first officer had radioed at 2046:03, "Yes sir, ah, we’ll be able to nold about five minutes, that’s all we can do," the airplane did not have sufficient fuel to fly to its alternate. Moreover, AVA052’s fuel state at the time it was cleared from holding at CAMRN to commence its approach to JFK was already critical for its destination. To help ensure sufficient fuel to complete a safe landing, an emergency should have been declared in order tu receive expedited handling. The airplane exhausted its fuel supply and crashed 47 minutes after the flightcrew stated that there was not sufficient fuel to make it to the alternate. This occurred after the flight was vectored for-an ILS approach to the destination, missed the first approach, and was unable to complete a second approach. 56 When the flight engineer entered 17,000 pounds of fuel remaining upon beginning the descent from FL370, the flightcrew should have estimated the distance and time remaining to destination, approach, alternate, and the reserve fuel required, in order to ensure that there was sufficient fuel for the flight. Included in these calculations should be the fue] quantity that the captain would want on board as he coimenced the first approach. This latter fuel figure, commonly referred to as the "minimum approach/landing fuel quantity," should be a part of a flightcrew’s calculations as the flight begins descent for landing. There is no indication that the flightcrew of AVAQ52 had calculated or established such a figure. Again, if a dispatch system had been functioning for AVA052, the dispatcher could have assisted in these calculations and contingencies could have been established jointly by: the dispatcher and flightcrew. The airline’s only written procedure for minimum fuel operation was published in its B-707 Operations Manual. The procedure was based upon an indicated fuel quantity in any main tank of 1,000 pounds or less. The procedure did not address a minimum fuel quantity for which a flight should be at the outer marker, inbound to the runway.

[ANALYSIS] 2.0 ANALYSIS 2.1 General The evidence confirmed that this accident occurred when the airplane’s engines lost power from fuel exhaustion while the flight was Maneuvering for a second instrument approach to JFK. Significant evidence was information contained on the CVR and examination of the airplane’s wreckage. There was an absence of fuel odor at the accident site, and no fire erupted during the impact. The only fuel found in the airplane was residual unusable fuel. There was no rotational damage to any of the four engines from impact forces, indicating that they had ceased operation before ground impact. In addition, the investigation found no engine or fuel system component malfunctions, including any that could have caused a premature exhaustion of fuel or a loss of fuel supply to the engine. The investigation revealed that the flightcrew had received the appropriate flight and ground training in the B-707 and that they possessed the appropriate flight and medical certification required by the DAAC of Colombia. Further, they had sufficient previous experience in conducting B-707 flights from Colombia to JFK. The investigation revealed that the airplane departed Medellin, Colombia, with sufficient fuel to fly to its destination. Its scheduled flight time was .4 hours 40 minutes. However, the airplane crashed 6 hours and 26 minutes after takeoff. Its flight plan called for a "required" fuel load of 72,430 pounds of fuel for the flight, which included fuel to JFK, reserve fuel, fuel to the alternate, and holding fuel, for a total of 6 hours and 8 minutes of flight. The evidence revealed that an additional amount of fuel was loaded aboard AVA052 above the required load. That amount was about 6,070 pounds, bringing the total planned load to about 80,500 pounds for takeoff at Medellin. The Safety Board believes that the actual fuel aboard the flight zi takeoff from Medellin was about 80,500 pourds, based on various documents evailable and the fuel figure notations made by the flight engineer during '- flight. Because of differences in the fuel numbers on some of these ents, this value could be in error either way by about 1,000 pounds; =heless, AVAO52 had sufficient fuel to complete the scheduled flight, as -¢ to meet other prescribed IFR fuel requirements. For example, there “ficient fuel to fly the scheduled route to the destination airport 52 (JFK), execute a missed approach, and fly to the alternate airport (Boston). However, Boston was forecast to be below IFR alternate minimums when AVA052’s flight plan was filed, and the actual weather there deteriorated further while the flight was en route.

[ANALYSIS] The tower controller did not follow up on the radio calls about running out of fuel. However, the TRACON controller turned the flight back onto a downwind leg and asked the flight'if it could accept a base leg 15 miles northeast of JFK. The first officer of AVA052 responded, "I guess so." Shortly thereafter, at 2124:22, the captain again advised the first officer to, "advise him we have an emergency." Four seconds later, the captain said, "did you tell him?" The first officer replied, “yes sir, I already advised him." Further, at 2125:08, the captain said to the first officer, "advise him we don’t have fuel." He asked again, at 2125:28, "Did you advise him that we don’t have fuel?" The first officer again said, "yes sir, I already advise him...." 58 These intracockpit conversations indicate a total breakdown in communications by the flightcrew in its attempts to relay the situation to ATC. The accident may have been inevitable at that point, because the engines began to flame out only about 7 minutes later. However, it is obvious that the first officer failed to convey the message that the captain intended. The evidence strongly suggests that the captain was unaware, at times, of the content of the first officer’s transmissions and that he did not hear or understand the ATC communications. The captain may have been preoccupied with flying the airplane and paying little attention to the first officer’s ATC radio transmissions. However, the Safety Board believes it more likely that his limited command of the English language prevented him from effectively monitoring the content of the transmission. The Safety © Board further believes that this deficiency might have been a factor in the accident, particularly if the captain believed that the first officer had adequately expressed the criticality of the fuel situation upon departure from CAMRN. . In summary, the Safety Board believes that the two key factors leading to this accident were the flightcrew’s failure to notify ATC of their fuel situation while holding at CAMRN in order to ensure arrival at the approach fix with an adequate approach minimum fuel level and a breakdown in communications between the flightcrew and ATC, and among the flight crewmembers. As a result of a fatal air carrier accident and an incident in which fuel exhaustion was determined to be causal, the Safety Board issued Safety Recommendation A-81-14 to the FAA on February 24, 1981.

[ANALYSIS] However, the differences are so diverse between the handling characteristics and emergency procedures of single-engine and multiengine aircraft, applicants for multiengine ratings who possess a single-engine instrument rating should be required to demonstrate their ebility to conduct safe multiengine operat: wns under actual or simulated instrument conditions, When an inflight emergency occurs, there is little time to decide the proper action to be taken. A preestablished plan of action and a thorough knowledge of the aircraft are requisites for the safe and efficient management of unusual, unexpected deviations from normal flight conditions, especially when the p.lot is burdened by the extra tasks associated with flight by instrument reference. The pilot had previously flown about 11 cross-country flights in NICOUV. His longest flight was 1 8/10 hours. The average time for each cross-country flight was about 1 1/2 hours, It is possible that the pilot may have flown elther all or most of these flights using only the main fuel tanks. Just as likely, however, is the possibility that he may have been accustomed to switching to reserve fuel further along in the flight, rather than shortly after leveling off. Since the amount of fuel onboard at takeoff, the fuel tank selection, the fuel distribution within the tanks, and the pilot's preferred procedures for takeoff, climb, and cruise are matters of conjecture, no significant conclusions could be drawn from the available fuel information. Additional evidence Indicates that the aircraft's engine(s) may have stopped because of fuel starvation. Witnesses reported hearing the aircraft making "popping" noises, The engine manufacturer indicated that an interruption of fuel flow to a Lycoming 10-720 engine could result in a popping noise or backfiring. At the first indication of abnormal engine operation, the pilot should have advanced the mixture, prop, and throttle controls to the full rich/high RPM/high manifold position. Having done so, he may have recognized his fuel inanagement error and attempted to correct it by turning the fuel boost pumps on and selecting auxiliary fuel. Thv surge or increase in engine power as described by witnesses may have been due to the resumption of fuel flow to the engines as a result of the pilot's selection of auxiliary fuel. The fuel tank selectors and boost pumps were located on the fuel control panel, which was located approximately 90° to the left and below the pilot's view of the primary flight instruments. Switching tanks would therefore have required the pilot to move his head down and to the left, thus diverting his attention from flying the aircraft. If, while under actual instrument conditions, the pilot's eyes were diverted from the flight instruments and his head was moved downward and turned (as when changing frequencies, checking flight log data, or changing fuel selectors), the aircraft rolled or turned at the same time and he suddenly returned his head to the normal position, a disorientation would most likely have occurred, A false sensation of diving or rolling beyond the vertical plane would have been produced. AS a result, there may have heen a strong, instinctive tencency to pitch or roll the aircraft in the opposite direction.

[FINDINGS] The Safety 3oard believes that this investigation indicates that certificate holders should have contingency pians for fighting very large fires such as fuel farm fires. l. 10. 55 CONCLUSIONS Findings Based on United’s refueling activity for the morning of November 25, 1990, and on the sequence in which pumps were to be activated, motor pump unit 3 would have been operating during the refueling of aircraft that were readied for flight between 0730 and 0900 mountain standard time. The motor on unit 3 gradually became loose during normal operations because of inadequately installed or maintained bolts, thus increasing lateral and vertical vibrations that caused overstress conditions on the bolts to the point of failure. Operation of the motor on unit 3 while it was no longer bolted down resulted in a lateral shifting of the motor relative to the punp by about 2.6 inches, which increased vibrations and friction that eventually fractured the pump case and two pump bolts and destroyed the coupler gears and the pump sealing surfaces. The destruction of the pump sealing surfaces and a fractured pump case resulted in a sizeable fuel leak because of the head pressure in the tanks that had been filled on the morning of the fire. fither the hot coupler or the motor on unit 3 provided an ignition source for the leaking fuel. Had tanks 3 and 4 been equipped with fail-safe control valves and internal fire valves with fusible links, the amount of fuel that fed the fire would have been significantly reduced, and. consequently, the duration and intensity of the fire would have been lessened. Had the control building been located in an area separate from the motor/pump equipment and outside of the containment area, vital records that would have been helpful to the investigation would not have been lost, and emergency response personnel could have accessed emergency shutoff switches and possibly could have remotely closed some of the contro? valves, thus reducing the amount of fuel that ultimately fed the fire. Had equipment thai «wonitors excessive temperatures and vibrations with automatic shutoff capability been installed on motor/pump unit 3, the equipment could have detected the vibration of the motor and shut down the unit, and the fire would not have occurred. Ti2 motor/pump equipment was not properly inspected for a substantial period of time before the fire. Management of AMR Combs failed to give proper priority to the task of inspecting and maintaining fuel farm pumping equipment. The airport certificate holder’s (city/county of Denver) lack of procedures or a contingency plan for responding to a fire of this magnitude prolonged the duration of the emergency. Federal Aviation Administration (FAA) regulations in 14 CFR Part 139 fail to specify the responsibility for inspection of fuel farms located cn airport property, when such fuel installations are operated by Part 12] and Part 135 air carriers.

[ANALYSIS] Fuel could have been leaking slowly from pump 3 for a period of time before the complete destruction of the sealing surfaces, which resulted in fuel being sprayed from the pump. [n an effort to determine more precisely the period of time during which the failure occurred and if fuel had been leaking for some time before the fire, dipstick records on sump tank 5 were reviewed. The records, wh'ch indicate the amount of fuel in the underground sump tank, showed that for about 2 weeks before the fire, there was a uniform increase of between 1.5 and 3 inches of fuel per day in the sump tank. However, fuel from a pressure relief valve in the Chase supply pipeline also emptied into this sump tank. Consequently, it could not be determined whether fuel frem this pressure relief valve or leaks from pump 3 were filling the sump tank. Further, the records indicated that there may have been some errors made in the dipstick measurements or the recording of the measurements. For example, the measurement for November 24, 1990, showed a 4-jnch decrease over the previous day’s reading, even though no fuel was removed from the sump tank. Consequently, the sump tank filling rate provided no insight into the period of time that fuel may have been leaking from pump 3 and, this, the period of time that the failure of the motor/pump unit 3 may have been occurring. In summary, although the damage inaicates that the failure of motor/pump unit 3 occurred over a period of time, probably days and possibly weeks, the precise time period could not be determined. The post-fire examination of motor/pump units 1 and 5 indicated that the motor and pumps in these units were also misaligned. However, there were no broken or missing bolts in either the pumps or motors. Consequently, the misalignment was determined to have been the restiit of warping of the bed plate from the intense heat of the fire. Duration aid Intensity of Fire Fuel leaking under pressure continued to feed the fire, and, as a result, the fire quickly intensified. The investigation examined the possible sources of the pressurized fuel leaks. 48 Computer records from Chase Transpo tation Company indicated tnat the fuel delivery on November 25, 1990, from its Aurora facility was terminated at 0925. The termination of pumping from Aurora was the result of a communicatica disruption, which suggests that the fire at the fuel farm damaged part of the communication system that controlled the pump at Aurora. Regardless of the reason for the disruption, the termination of the pumping operation from Aurora indicates that fuel was not being supplied to the fire by the oump in Aurora. However, when the communication interruption occurred, the pump stopped and a valve closed at the discharge side of the pump in Aurora, maintaining the pressure on the pipeline at about 220 psig. The motor-operated supply valve at the fuel farm that directed fuel to Continental’s tank 7 remained open because of the communication interruption.

[FINDINGS] The airport certificate holder’s (city/county of Denver) lack of procedures or a contingency plan for responding to a fire of this magnitude prolonged the duration of the emergency. Federal Aviation Administration (FAA) regulations in 14 CFR Part 139 fail to specify the responsibility for inspection of fuel farms located cn airport property, when such fuel installations are operated by Part 12] and Part 135 air carriers. Similarly, FAA regulations in 14 CFR Parts 121 and 135 do not directly aduress the inspection of fuel farms operated by air carriers. The airport certificate holder did not allocate sufficient resources to perform thorough quarterly inspections of fuelers on airport property. Probable Cause The National Transportation Safety Board determines that the probable cause of the fire at the fuel storage facility at Denver’s Stapleton International Airport was the failure of AMR Combs to detect loose motor bolts that permitted the motor of motor/pump unit 3 to become misaligned, resulting in damage to the pump and subsequent leakage and ignition of fuel. Contributing to the accident was the failure of AMR Combs to properly train its employees to inspect and maintain the fuel pump equipment and the failure of the city and county of Denver to carry out its certificate holder responsibility to oversee the fuel storage facility in accordance with its airport certification manual.

[(a) FINDINGS] We are also aware of FAA's current proposed rule changes which, if adopted, would -:esult in improved and updated airworthiness standards applicable to airccsft performance, flight characteristics, flight manuals, and operating limitations and information. The Safety Board believes this accident illustrates further the need for early accomplishment of these measures to insure that this critical information is available to all piiots. 2.2 Conclusions (a) Findings 1. Neither wrather nor air traffic control were factors in this acciderc. 2. When the aircraft was between 50 and 100 feet above the runway, its left engine fatled because the crankshaft broke at the No. 7 crankcheek. The crankshaft broke because a pre-existing fatigue crack had propagated through the crankcheek and the cheek could no longer withstand ncrmal operating stresses. No material defects were found in the crankshaft that could have contributed to the initiation of the crack. The reason for the initlation of the crack was not determined, but it probably was initiated by a combination of residual stresses and operating stresses. The right engine continued to function normally after the left engine fatled. According to performance. charts and flight tests, the aircraft was cupable of climbing about 300 fre at 117 mph witht one engine inoperative, its propeller feathered, and the landing gea™ retracted. According to performance charts and fligh* tests, the aircraft could not maintain level or climbing flight in the takeoff configuration with one engine inoperative and its propeller windmilling, and a descent of 100 to 200 fpm was required to maintain te: soff airspeed. ‘a, {{OV Ai emcee carpe. sasha rccmneniinatinie int OO cry pase tse amyens:, tts CA tail tet e Following failure of the left engine, the pilot lost control of the aircraft because he permitted the airspeed to decrease below Vee, and he permitted the aircraft to stall. The pilot did not retract the lending gear either before or after the engine fatted. The left propeller probably feathered automatically after the camshaft broke; however, until the propeller blade angles had increased substantially, the propeller would have windmilled and would have produced significant drag. The pilot's ability to maintain flight after the engine failed was marginal beceuse of nix rime conditions, the aircraft's low altitude, and the aircraft's inability to climb or maintain lev-1 flight in the takeoff configuration with one engine inoperative and its propeller initially windmilling. The accident was not survivable because of the high impact forces generated by the high rate of descent and the fire that erupted after impact. (b) Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the sudden failure of the airplane's left engine at a point on the takeoff flightpath where the airplane's single-engine performance in the takeoff configuration anu {{ts height above the ground combined to make the pilot's ability vo sustain flight marginal.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] The Safety Board's metallurgical tests disclosed no material defects in the crankshaft, nor did the manufacturer's testing produce a probable operating stress that could have contributed to the initiation of the fatigue crack. Therefore, the cause of the crack's initiation was not determined. However, because ne metallurgical defects existed, we believe that the crack probably was initiated by stresses inherent and applied; that is, by a combination of residual stresses (which probably were introduced sometime during the manufacturing process) and normal or cbnormal operating stresses. Although major repairs had been made to the left engine abuut a year before the accident, there is no evidence that these repairs affected the condition of the crankshaft. This is because the fatigue crack originated from a subsurface crea rather than an area that might have been exposed and damaged during the course of the repair work, The right engine continued to function properly after the left engine failed. The ccndition of the right propeller indicates that it was turning at a relatively high rpm at impact but was not under full power from the engine. This condition is consistent with the pilot's apparent retention of lateral and directional cuntrol of the aircraft, whicn struck slightly right-wing low, because he probably reduced power or the rigiit engine sometime shortly before tryact in an effort to control the aircraft laterally and directionally. -~ 15 - The exact time at which the crankshaft broke wars not determined. However, since the landing gear apparently remained extended until the aircraft struck the ground, the crankshaft probably broke before the pilot normally would have retracted the gear, or when the aircraft was somewhere between 50 and 100 feet above the runway. Consequently, che pilot probably was distracted from tnitfally retracting the gear hy the failure of the left engine. After the engine failed, the pilot apparentiy did not perform the first item in the "Engine Failure After Lift-Cff" procedure because the landing gears were extended in the wreckage. It is not known why he did not retrac® the gears, but under the circumsrances, and where insufficteat runway remains on which to land the aircraft, an almost instinctive reaction might have occurred: An immediate atterpt to seek and correct the cause of the perceived engine malfunction and, in the process, to increase the aircraft's altitude by increasing the rate of climb at the expense of any ecess airspeed. It is possible that this pilot reacted in such a@ manner. (n the other hand, considering (1) the possible elapsed time between engine failure and impact, (2) the rapid degradation tn airspeed that occurs after loss of an engine if the landing gear are not retracted, the propeller feathered, or the pitch attitude decreased, and (3) the single-engine stall speed of near 90 mph, the pilot had little time or altitude in which to recognize the problem and respon? with the appropriate corrective action.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] This conclusion is confirmed by the short distance the rircraft traveled after impact, the crushed and flattened unterside of the fuselage and engines, and the impact damage to the pilots" and passengers‘ seats. The Safety Board is concerned thet such an engine failure can be catastrophic. We believe that 13 fractures in the same area of the No. 7 crankcheek for undetermined reasons are sutficient to indicate that a problem exists with ttis particular crankshaft. Because the reason for the initiation of -ne futigue crack probably involved a combination of residual stresses and operating stresses, we believe it important that the wanufacturer continue its efforts to identify and eliminate the causs: of the crack initiation. Yiese efforts should include a complete analysis of residual stresses introduced during the manufacturing process. Furthermore, the Federal Aviation Administration should initiate a program: (1) For inspecting the crankshafts from I10520 series en;zines for cracks whenever the crankshafts are available for inspection, ana (2) for reviewing the results uf these inspections to determine whether any deficiencies exist in chese engines. However, because engines fail for many reasons, including a substantial number involving operator errcer, and to fur her reduce the possibility that catastrophic accidents of this kind will be repeated, we believe that improvements are needed in the information and training on single-engine performance that is provided to pilots of light twinengine airplanes. Most pilots of these aivplanes are probably aware that, under some combinations of gress weight, density altitude, and airspeed, the airplanes can maintain level flight or climb in the takeoff configuration with one engine inoperative and the propeller windmilling. These generally are tine conditions which pSlots are exposed to in training, initial checkouts, and flight checks. However, they may or may not know that under other conditions, the airplane will be unable to climb or maintain level flight in the takeoff configuration with one engine inoperative and its propeller windrilling. This is particularly true for high gross weights and high density altitudes--conditions that the average pilot is not exposed to in training. Consequentiy, a careful study and analysis of the =irplane'a performance charte, if available i, is required to deternine the airplane's single-engine capabilities for a given set. of conditions. Many pilots night not fully appreciate the its significance of Vac 865 related to aircraft stall speeds with takeoff power on one engine. As shown in the Baron 58 flight tests, the single-engine st3l' speed in the takeoff conf-guration was near 90 mph, cr quite close to the certificated Vac Of 94 mph. Also, although the test aircraft was controllable at 90 mph, it is likely that iateral and dircctional controi beiew Vy, would be difficult to maintain with full power on one engine the lower density altitude existing at the time of the accident.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] Additicnally, the nighttime conditions probably would have made both recognition and response more difficult. If the aircraft entered the stall regime at 150 feet above the runwey, an accident would have been unavoidable. The left propeller apparently was at or near the feattered position when the aircraft struck the ground because: (1) The propeller was rotating slowly or was not rotating at all, (2) the chord line, near ground level, of the partially buried propeller blade was approximately aligned with the aircraft's impact heading, and (3) there was vo apparent twisting of the blade in the ground after it had been buried i:*~ the ground. Although the left propeller biades were founc at a pi:..: angle of about 21°, the Safety Board believes that the aircraft's left rotation after impact probably rotated the blades frow a near feather angle (about 80°) to contact with the centrifugal responsive pins before the propeller broke from its shaft. In fact, the partially buried blade probably acted as an anchor which, coupled with che aircraft's forward momentum, produced the left turning inoment on the aircraft after impact. Consicering the critical conditions with which the pilot was faced following failure of the crankshaft, and since he did rot perform the first item in the engine failure procedure, the Safety Board believes that the left propeller probably feathered automatically after the ~ 16 = camshaft broke and disrupted the oil supply to the propeller governor. This disruption probably occurred several seconds after the crankshaft failed, and, during this time period, the propeller would have windmilled and would have produced a significant amount of drag. When combined with the extended landing gear, the aircraft's climb capability would have been reduced sustantially--possibly to as little as a negative 109 fpm at 1]? mph. If the airspeed was less than 117 mph, the rate of aescent could have been greater. As the propeller moved toward the feather position, it would have produced less drag, which eventually should have enabled the pilot to establish a climb even with the landing gear extended if he had maintained the appropriate airspeed. However, since the aircraft struck the ground with considerably more force than would have been generated by a 100 fpm rate of descent, the Safety Board concludes that the pilot stalled the aircraft in an attempt to maintain altitude and avoid the ground. This conclusion is confirmed by the short distance the rircraft traveled after impact, the crushed and flattened unterside of the fuselage and engines, and the impact damage to the pilots" and passengers‘ seats. The Safety Board is concerned thet such an engine failure can be catastrophic.

[(b) PROBABLE CAUSE] (b) Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the sudden failure of the airplane's left engine at a point on the takeoff flightpath where the airplane's single-engine performance in the takeoff configuration anu {{ts height above the ground combined to make the pilot's ability vo sustain flight marginal. The pilot's failure to retract the landing gear and control the airplane to maintain a safe airspeed contrituted to the accident and wers. factors in causing the high acceleration loads when the airplane struck the ground. RECOMMENDATIONS As a result of this accident, the National Transportation Safety Board has recommended that the Federal Aviation Administration: RR arrears 2 reine rere nn i *, & aE a’ 3 von and repair faciJities to inspect the 10-520 series crankshafts for incipient or develc ped cracks, preferably using an inspection eans capable of detecting subsurface cracks, in the vicfnity of the short trankchecks any tine that the cranksta‘ts are available for inspection. (Class II-Priority Followup) (A-77-- 43) J "Tssue a maintenance alert bulletin to advise engine overhaul "Conduct a directed safety investigation consisting of a review of overhaul and repeir facility inspection results to determine if the frequeacy and distribution of detected fatigue cracks indicates a deficiency in the 10-520 engine. (Class Il-Priority Followup) (A-77~44)" BY THE NATIONAL TRANSPORTATION SAFETY BOARD. /s/ WEBSTER B. TODD, JR. Chairman /s/ KAY BAILEY AR nen ee cmeamiuamiesmantane Vice Chairman /3/ FRANCIS H. McADAMS Member /s/ PHILIP A. HOGUE Member WILLIAM R. HALEY Meaber June 16, 1977 oy eg ie, SE NINN. “LER RRe Ln Mn ORN NMRA? AANA LemmeniRtane aA amet yori: el, EM IT tang Y + amie, en ath bane wn * oe mee Sth eeeaenidantatetiiotemns.s aamenied italia’ kat MEE a pgs APPENDIX A INVESTIGATION AND HEARING 1. Investigation The National Transportation Safety Board was notified of the accident about 2145 on August 3, 1976. Investigators frum the Kansas City Yield Office proceeded immediately to the scene, Additional investigators from Washington, D.C., were sent later. Investigative groups were established for structure 1/systems and powerplants. The Investigator-in-Charge was responsible for operations, air traffic control, witnesses, human factors, and maintenance records. --- Footnotes: [3/ ° +14 CFR 135, Special Federal Aviation Regulation 23.]

[PROBABLE CAUSE] The flight returned to the airport after shutdown of the engine and extinguishing of the engine fire. "During the return to the airport, the flightcrew experienced difficulty in extending the landing gear and the wing flaps after parts of the feiled engine severed the hydraulic and pneumatic systems' supply lines. The captain elected to "go around" and extended the landing gear by the alternate system. The aircraft made a successful landing, and there were no injuries tc the 15 crewnembers or the 127 passengers. “Our preliminary investigation of the engine failure revealed that a separation occurred to the rim portion of the second-stage turbine disk. It has been confirmed that failures of at least four of seven first-stage turbine slades contributed to the fracture of nunerous second-stage turbine vane fect. As a result of the cumulative effect of the broken vane feet, an aft deflection of the nozzle support resulted, causing interference with and rubbing of the second-stage turbine disk. Progressive weakening of the disk rim area resulted in the in-flight failure of the rim. We have also confirmed that although failure mode of this second-stage turbine disk rim was similar to that of the Air France JT9D-3A engine failure of August 17, 1976. the failure mechanism was entircly different. “As a result of our investigation and meeting with Pratt & Whitney engineering staf® personnel and your Eastern Region Flight Standards personnel, immediate inspection action was initiated. This was considered fully responsive to the immediate needs of this situation. The Safety Board commends the Administrator's formalizing this corrective action in the form of your engineering alerts of September 19 and 23, 1970. "In view of the potentially catastrophic results of the failure such as experienced by American Airlines, the Board remains concerned about this matter in the longer range sense and would urge the Administrator to initiate further expeditious actions in order to preclude recurrence of similar failures. Accordingly, the Board offers the following observations. "It is generally recognized that the JTQD engine is normelly operating near critical turbine temperature conditions. This is particularly crue when operating in high ambient temperatures. Several JTQ9D engines have recently been removed from service and returned to Pratt & Whitney for overhaul, because cf failed firststage turbine blades as well as broken second-stage vane feet. There is evidence that these failures had occurred as the result of operation at higher-than-desirable temperatures. "In the case of the most recent American Airlines turbine disk rim separation, there was evidence that at least six firststage turbine blades had sustained varying degrees of fractures some time prior to the final failure. Our technical staff finds it most difficult to reconcile the fact that the airborne vibration monitoring equipment installed in the aircraft was either inadequate or was not effectively utilized in detecting this condition.

[ANALYSIS AND CONCLUSIONS] The engine normally operates at relatively high turbine temperatures and therefore requires most precise monitoring of all vital operating parameters and effective analysis of any confirmed deviations from normal parameters, Vibratory stresses in exzess of four times their normal level were imposed upon the second-stage vane feet after stress rupture of first-stage turbine blades. Multiple failures of second-stage vane feet and resultant rearvard shift of the nozzle inner support caused rubbing of the support against the second-stage turbine disk until separation of the disk rim occurred, Deterioration of vital engine operating parameters was evident on both the narrative portion of the flight log and the computerized engine condition monitor log. Maintenance actions taxen by the carrier in attempting to correct the in-flight discrepancies as reported on the No. 1 engine were not responsive to the problem that existed. While the computerized engine monitor log used by the carrier was effective in accurately identifying the progressive decrease of N2 and increase in EGT and fuel flow, the data was not available for use in time to effect corrective action prior to severe engine failure, The fire which resulted from the turbine failure was terminated by the immediate response of the flightcrew in sucecssfully shutting off fuel supply to the No. 1 pylon. The fire extinguishing agent appeared to have little effect in conbating the fira. The results of the primary failure affected other systems critical to the landing phase and compounded the already existing emergency by placing an additional burden upon the flightcrevw. PROBABLE CAUSE Tne National Transportation Safety Board determines that the probable caus: of this incident was e progressive faflure in the high-pressure turbine sodule in the No. 1 JTYD-3\ engine. This failure was initiated by the undetected stress rupture fractures of several first-stage turbine blades and culminated in in-flight separation of the second-stage turbiuie disk rim. -lRECOMMENDATIONS AND CORRECTIVE ACTION On September 25, 1970, the Board sent the following letter to the Adninistretor of the Federal Aviation Administration: “The National Transportation Safety Board is now investigating the JT9D-3 engine failure and in-flight fire involving American Airlines, Boeing 747, N743PA, which occurred during takeoff from the San Franetsco International Airport on Septemher 18, 1970. A failure occurred in the No. ] engine 13 seconds after lift-off, followed by a fire warning. The flight returned to the airport after shutdown of the engine and extinguishing of the engine fire. "During the return to the airport, the flightcrew experienced difficulty in extending the landing gear and the wing flaps after parts of the feiled engine severed the hydraulic and pneumatic systems' supply lines. The captain elected to "go around" and extended the landing gear by the alternate system.

[PROBABLE CAUSE] There is evidence that these failures had occurred as the result of operation at higher-than-desirable temperatures. "In the case of the most recent American Airlines turbine disk rim separation, there was evidence that at least six firststage turbine blades had sustained varying degrees of fractures some time prior to the final failure. Our technical staff finds it most difficult to reconcile the fact that the airborne vibration monitoring equipment installed in the aircraft was either inadequate or was not effectively utilized in detecting this condition. We &lso feel that other engine instrumentation, namely: fuel flow, engine pressure ratio, and exhaust gas temperature should have been capable of collectively reflecting appropriate changes in the engine's operating parameters, if such instrumentation were properly calibrated and the respective readings were recorded and closely analyzed, "In this area, we recommend the following be ccnsidered, 1, Initiate appropriate action toward the operators’ maintaining a program of current engine condition monitoring. Review engine instrumentation calibration and existing instrument tolerances to assure the most precise engine operating parameter indications. "Further, it appears that the reliability of the Boeing 747 auxiliary power units is somewhat marginal. When engine starts must be accomplished by the use of ground units, pneunatic duct - pressures may often be less than what is required, even when multiple units are used, The result is usually a start that may involve a temperature rise, approaching the “recoverable stall” condition. Since exhaust gas temperature, although above normal under these conditions often do not exceed the published limits, no record is made of these occurrences, and there is no possible way to determine how many times an engine hot section has been exposed to higher-than-normal temperatures, The effects of thermal transients are known to be cumulative and conceivably affect turbine vlade reliability. ~-19- "As another measure toward improving the service reliability of first-stage turbine blades, it is recommended that appropriate action be initiated to: 1. Improve the reliability of auxillary power units in order to reduce the probability of high thermal transients while starting engines with marginal air supply. Ensure that flightcrews maintain adequate pneumatic air duct pressure during engine starts. Record any abnormal starts when an approach to 4 “recoverable stall" is experienced. Establish precise limitations regarding the number of "approaches to recoverable stall" conditions which may be tolerated without cumulative adverse effects upon turbine blade durability. "The Safety Board is aware that the manufacturer has developed an improved type first-stage turbine blade (vented) which is expected to provide improved cooling characteristics and be more reliable when operating at high temperatures. "With respect to the improved first-stage turbine blades, the Safety Board recommends: 1. Incorporation of the “vented” first-stage turbine blade in ali JT9D series engines be the subject of regulatory action as soon as sufficient production is assured and service bulletins and engineering orders are formulated by the manufacturer. "Water tnjection is presently being used on an optional basis by individual operators. Since water injection allows utilization of 45,000 pounds of thrust versus 43,500 pounds for take-off, some operators elect to use water only when takeoff weight, runway lengths, and ambient temperature conditions require the maximum thrust rating of 45,000 pounds.

[ANALYSIS AND CONCLUSIONS] These were not designed to withstand the nearly 4.1 increase in vibratory loads which occurred when the initial failure of the firststage blades occurred, The fatigue failures of these inner shroud near vane feet allcwed the second-stage turbine stator shroud assembly to shift rearward and rub against the front web surface of the second-stage turbine hub. This continued rubbing caused grooves to be worn into the disk portion of the front turbine hub. Finally, the front turbine hub fractured circunferentially in a rapid tensile manner at the grooved and weakened points, relating the entire rim porticn of the disk. {{ In reviewing the operating history of the engine (S/N 662274), it @ becomes quite apparent that the laboratory findings relative to the pre-existing failures of at least four first-stage turbine blades is quite accurate. -32.- Flight discrepancies reported on three separate occasions, indicating “poor throttle alignment" and, in one case, the No. 1 throttle's being as mucr as 3 inches forward of other engines, reflected an apparent deterioration in engine performance. However, due to the chronic throttle alignment problems which are, according to the carrier, normally encountered on the 747, as well as on other airplanes, this symptom was improperly diagnosed. The corrective action, consisting of rig check and adjustment of No. 2 thrust lever as well as swapping of No. 1}} and No. 2 EPR indicators, was however, responsive to the flight discrepancy as reported. The flight engineer hed obviously attempted to describe the cause of the problem by stating "No. 1 and No. 2 throttles very far out of rig,” rather than to describe accurately the symptoms which possibly would have required more intense troubleshooting on the ground. A subsequent flight discrepancy reported on September 17, 1 day prior to the final failure, showed the No. 1 engine to be EGT limited on takeoff and climb. It was necessary to operate No. 1 engine at .09 EPR less than others to maintain an &0T of 775° F. Other engines on the airplane, during this phase of the flight, operated at 1.31 EFR, maintaining 766° F. A thrust reduction of .09 EPR at takeoff under standard conditions can be translated into approximately 6,500 pounds of thrust and 105° F. of EGT, Here again, it is apparent that the corrective action in replacing the EPR transmitter was based upon an erroneous assumption that an instrument error wos responsible for the low-thrust indication on the No, l engine, An effective troubleshooting program at this point would, in all probability, have determined the reason for the high EGT and low EPR. Although tailpipe inspections were performed after every flight, the evidence of an incipient failure was either not present or was not recognized. The inputs into the computerized JTQ9D condition monitoring program for this engine from September 10 through 16 likewise reflected adverse changes in trends of vital operating parameters which were indicative of the possibility of a serious engine malfunction.

[PROBABLE CAUSE] Instructions on engine starting have been reviewed with the carriers at several of our industry meetings. Adherence to these procedures should prevent hot starts. "Fhe suggested mandatory use of water injection on ali aircraft on takeoffs is not viewed as a panacea to the turbir blade-cracking problem as turbine blade problems have not been confined to "dry" eigines, The use of water offers a reduction in turbine gas inlet temperature only for moderate ambient temperature takeoff conditions. More effective in this area is the procedure of using reduced thrust levels where possible to lower the turbine gas temperature. "The incorporation of improved turbine blades is considered to be the pest solution for improving the durability of these parts. First-service use of "vented" turbine blades has begun. Limited quantities of the new type "vented" turbine blades are avatilable new and are being installed as rapidly as practicable. The carriers are estimating completion of retrofit on their fleet engines in the latter half of 1971. "Pratt & Whitney Aircraft is developing further improved turbine blades of a more heat-resistant material, These should be available in the near future, "It was agreed on 2 October 1970 that borescope inspection frequency of the ccmbustion section would be established at 100-hour intervals on engines with more than 500 hours or more than 250 cycles. This, we believe, will effectively detect blades cracked from heat distress before they progress to fatlure, " ~ 20. CORRECTIVE ACTION Tne manufacturer, in conjunction with FAA advisories, has require?: 1, Radioisotope inspection program for a11 JT9D engines which is in effect to inspect for vane lug failures. In addition, the following inspections are required: Inspect engine tailpipe for metal after arrival at each station; Monitor aireraft airborne vibration monitoring equipment; If there is an indication as a result of inspection (a) or (b) above, borescope or chamberscope the turbine area for failed first-stage turbine blades; and Continue borescope inspection of the turbine area on & scheduled vasis, but in no case exceeding 200-hour intervals. In addition to the above field inspections, the following engineering change improvements are either being incorporated in the JTOD engine now or are scheduled for incorporation within the next several months: Revised first-stage turbine blade leading edge cooling plus improved blade material. Impingement-cooled first-stage turbine blade. Second-stage vane rear inner lug thickness increased from .085 to .110 inches. Increased cooling flow to second-stage inner seal and support , ard second-stage vane lug.

[ANALYSIS AND CONCLUSIONS] Although tailpipe inspections were performed after every flight, the evidence of an incipient failure was either not present or was not recognized. The inputs into the computerized JTQ9D condition monitoring program for this engine from September 10 through 16 likewise reflected adverse changes in trends of vital operating parameters which were indicative of the possibility of a serious engine malfunction. These data points showed a progressive increase of BGT and fuel flow while Ne showed a progressive decrease over the same period of time. Due to the time lag between data acquisition, the computer printout, and analysis of this above data, the trends shown were not available until after the failure had occurred and, consequently, could not be used effectively in diagnosing the problem. The difficulty in starting the No. 1 engine, which was experienced at the origination of the flight, is not necessarily considered a major factor in this particular failure. It is noted that the airplane's APU was inoperative when N743PA was released for flight. While the APU is needed for ground operations only, its functions, such as supplying pneumatic pressure for engine starting, are vital. In order to obtain a satisfactory engine start without the risk of approaching overtemperature conditions, a minimum of 30 p.s.i. pneumatic duct pressure should be available. The APU is capable of supplying between 40 and 45 p.s.i. pneumatic pressure; on the other hand, ground air supply units, which must be used when the APU is inoperative, generally do not have this capability unless the dual or triple external air connections on the airplane can be supplied by the ground air units. In summary, the probability of an overtemperature condition during engine starting is emsidered higher when the airplane's APU system is inoperative and marginal capab/lity ground units are utilized. - 7h - EFFECTS OF ENGINE STRUCTURAL SAILURE UPON THE ATRCRAFT In view of the compounded and direct effects of a turbine failure such as this upon the continued safe flight of the airplane, the Safety Board finds a need for reviewing this aspect of the occurrence. As indicated by the crew of Flight 14, the natural and most immediate concern was the control of the fire in flight and the safe return to the airport. Although the extinguishir, agent was discharged by the crew, the agent's effectiveness in controlling or extinguishing the ftre seems quite questionable. The most serious impairment of the system's effectiveness occurred when the engine and pylon enclosures were penetrated during the turbine failure, allowing a substantial portion of the extinguishing agent to escape into the atmosphere. The fire inside of the pylon continued with such intensity that both of the agent containers became physically detached from their mountings and fell to the bottom of the pylon structure. This fact alone can leave little doubt that the fire continued for some time after the agent was discharged.

[ANALYSIS AND CONCLUSIONS] The fire inside of the pylon continued with such intensity that both of the agent containers became physically detached from their mountings and fell to the bottom of the pylon structure. This fact alone can leave little doubt that the fire continued for some time after the agent was discharged. It is the opinicn of the Board that the fire terminated only when the sources of flammable materials became exhausted, In this respect, termination of fuel supply was effective since the fuel line was severed downstream of the firewall shutoff valve which was closed by the timely action of the flignterew. Flame propagation over both the top and bottom of the wing was such that there was danger of ignition of the fuel which was leaking out of the punctured fuel tank access plates. In the ease of the No. 1 hydraulic system supply, line severance occurred between the reservoir and the shutoff valve, allowing depletion and Jeaxage into the fire area of the total fluid supply for the No. 1 system. Fluid supply was then no longer available to the No. 1 system's air-driven hydraulic pump. This pump normally provides a backup pressure source for the No, 1 system in case engine pump pressure is either lost or derands upon it become excessive. Of further significance is the puncture of the left wing pneumatic duct which supplies pressure for all of the pneumatically operated units in the left wing. Consequen3ly, the operation of the No. 2 air-driven hydraulic pump would have been impaired by greatly reduced pneumatic pressure, if such operation became 4 requirement. The loss of No. 1 hydraulic system, requiring ali ernate extension of the body landing gear and leading edge flaps, caused a delay in returning to San Francisco Airport and placed an additional burden upon the flightcrew during the already existing fire emergency. Although other, less vital systems were either fully or partially deprived of normal hydraulic and/or pneumatic pressure, there appeared to be no further adverse effects upon the operation of the airplane. CONCIUSIONS From the investigation of this incident, the Safety Board concludes the following: 1. There were no material deficiencies of the HPT module which either caused or ccntributed to the failure. 2. The engine had been allowed to attain operating temperatures which were sufficiently in excess of design limits to initiate stress rupture failures of first-stage turbine biades. The engine normally operates at relatively high turbine temperatures and therefore requires most precise monitoring of all vital operating parameters and effective analysis of any confirmed deviations from normal parameters, Vibratory stresses in exzess of four times their normal level were imposed upon the second-stage vane feet after stress rupture of first-stage turbine blades.

[ANALYSIS] There are no alternate mean for the cngine to receive fuel. 2. ANALYSIS AND CGNCLUSIONS 2.1 Analysis There was no evidence of structural failure, malfunction, or abnormality of the airframe, contro! systems, powerplants, and other comnponents other than the failure of the No. 1 engine. The failure of this engine occurred at a height of approximately 50 feet and above V4 speed. Failure of this engine would not have caused the accid’ .t, as the aircraft at the time of the engine failuse was capable of continuing to climb on one engine and to make a subsequent safe landing. The only causal factors involved in the accident were those directly associated with the powerplants and the operational procedures used by the crew. In assessing the powerplant factors involved, it was confirmed that the Ne. 1 engine was not rotating at ground contact. The cause of the inflight failure of the engine was a M082 heatcode first-stage turbine blade failure. This blade failure is typical of other M082 heatccde failures ia that the blade material contained a concentration of 1.6 parts per million of che cramp eletacnt bismuth, Laboratory examination of previous M082 heatcode blade failures <-eclosed concentrations of 1.4 co 1.9 parts per million of besmuth. The airline operators and the engine manyfacturer are cognizant of this problem. The engine manufacturcr has recommended that the first-stage turbine blades be examined at the next heavy maintenance and that all blades identified with heatcode M082 be removed from setvice. To accomplish this, United Air Lines initiated a prograny on March 13, 1970, to identify and remove these blades from service. United had “gg west! eT a . I examined more than 260 engines in the implementation of this program. The No. 1 engine was scheduled for examination at the next heavy maintenance check, which would notmally have occurred at 5,800 hours or less. Ie is apparent the No. 2 engine was in an operable condition at the time of the accident. All tie evidence reveals the engine was operating in the aie, during the thrust reversing cycle, and until the aircraft came to rest. Results of examination and testing of the fuel pump drive shaft showed that the fracture was caused by fatigue resulting from a rotating beam type loading. The ioading induced on this shaft was due to an extreme misalignment between the fuel pump and the spur gear drive shaft. tt was estimated that che misalignment created a bending stress in excess of 180,000 p.s.i. in the fucl pump drive shaft. Such a high bending stress results in ratchet marks, indicating multiple fatigue origins, which increase in number with the degree of stress concentration and stress level. The number of cycles to failuce is primarily dependent on the level of this bending stress. In this case, the number of cycles is assumed to have been low.

[(a) FINDINGS] 2.2 Conclusions (a) Findings 1, The fligatcrew members were properly certificared for the flight. Zz. The aircraft was properly certificated and airworthy, 3. The weight and balance of the aircratt was within che allowable limits. 4, Ac the gross weight et which the aircrafe was being operated, it was capable of climbing on one engims. Rm Ew y aay §. The No. 1 engine failed in flight as a a result of a first-stage (Nz) turbine blace failure. 6. The No. 2 fuel pump drive shaft faiied at impact of the engine. aN + 7. The No. 2 engine was operating until rao \ it impacted the ground, oy] 8. The aircraft was airborne and above 1 oa IB V> speed at the time of the engine fail- rao ure, ae 9, The flightcrew did aot properly We I utilize the engine and aircraft instru- ae ments to determine the condition of the F. engines, alticude, and airspecd. 10. Company procedures and applicable i. flight manuals dictate that the flight : should have been continucd with one iy a F ¢ * a Ve Ug engine inoperative. e I \ \ Fi '"s ma 11. The captain discontinued the takeoff and landed back on the runway. i J 12. The captain erroneously decided power to both cngines had been lost. selected and power was applied after ii 13. ‘fhe No. 2 engine reverse thrust vas ‘ touchdown, 1 J 14. The captain had satisfactorily 4 accomplished an engine-out takeoff in the simulator and two a the aircraft since March 12, 1968. me 15. The first officer remained on the I controls after the captain took over the f a, control of the aircraft. , I fA fb) Probable Cause ; I 3g The National Transportation Safety I - Board determines that the probable cause of this accident was the termination of the takeoff, after the No. 1 engine Failed, at a speed above V» at a height of approximately 50 feet, with insufficient runway remaining to effect a sate landing, The captain's decision and his action to cerminate the takeoff were based on the erroncous judgment that both engines had failed. 3. RECOMMENDATIONS During its deliberations, the National Transportation Safety Board found that important safety lessons were evident from the facts, conditions, and circumstances of this and similar accidents. The Board, therefore, recorimenils to the Fee-ral Aviation Administration the fol- lowing: 1. Reassess the respective duties and responsibilities of the captain and the first officer during critical pases of flight.

[ANALYSIS] Since the captain did no¢ note both sets of engine instruments spooling down, and since the left engine svffered a tapid total power failure, ane must conclude thar he read the left set cf instruments and interpreted them to be the right set. Ie can be assumed then. that a tight yaw, observed by the captain, which may have induced hiin to transpose the instrument readings to be compatible with a yaw to the right. Ie is interesting to note that the first officer's impression was that the left engine was “spooling down,” Perhaps the first officer had a better feel for the alreraft prior to the captain's assuming control. if such a misinterpretation as two which engine had failed remained fixed, subsequent subjective “feel" for the aircraft could have been confusing. In this accident, accurate “feel” for the aircraft may also have been compromised by the presence cf both pilors on the controls. The captain stated he applied full power with no tesponse in airspeed, engine noise. or udder fee!, Very little increase in thrust would have Be) resulted from this action. The increased power lever input would not have given significant increases in the parameters of response for which the captain was looking. All three crewmen stated there was a steady and substantial loss of engine ncise before she attempt to land was initiated. The decrease of engine noix associated with the loss of one engine, located symmetrically with the second engine in relation to the flight deck, could be perceived as no mote than a 3 decibel decrease. This decrease, if noted at all, would not be alarming. It also would not have bcen perce ved as agradual steady loss. It is difficult to explain the reason the flight deck crew heard a steady and substantial decrease in engine noise, One possible explanation is that there is a substantial decrease of engine noise in the cockpit when an aircraft leaves the ground on takeoff. The captain's assessment of the emergency was chat the aircraft would not sustain flight. He was then forced to make en immediate decision as to where to make the inevitable landing. Since a portion of the runway was stil visible, his choice was to land back on the runway. Fucthermore, the need for a rapid decision in order to effect a return to the runway greatly compromised the time available co assess the emergency. 2.2 Conclusions (a) Findings 1, The fligatcrew members were properly certificared for the flight. Zz. The aircraft was properly certificated and airworthy, 3. The weight and balance of the aircratt was within che allowable limits. 4, Ac the gross weight et which the aircrafe was being operated, it was capable of climbing on one engims.

[ANALYSIS] At fucl shutoff, the percent of No dropped off to 41 percent in a maximum of 8 seconds. If the aircraft did have a double engine failure, the normal electrical generating systems would have been lost. Electrical power on the aircraft was available throughout the overrun. The only question is — was it coming from a normal bus or from a standby bur. and the airceafe battery. One trans. fer of ¢': ical power was evidenced on the voice re. tf, This would have occurred when the No. 1 engine failed, since the normal power source for the voice recorder comes from the No. 1 radio bus. There was no evidence on the Right recorder of a power loss and it is powered from the same bus. The fact that the flight and voice recorders operated throughout she overrun is conclusive evidence that electrical power from a normal geherating system was available. The flight and voice recorders cannot be powered from the standby buses. In addition, ac the landing gear on the aircraft was down, the electrical circuie for the flight recorder could not have been completed unless an engine was running. The electrical circuit for the flight recorder is completed through either engine oil pressure switch or the landing gear latch relays. The similarity of the compass information on the pilot’s and copilot’s Course Indicator (Cl) and Radio Magnetic Indicator (RMI) instruments would indicate that normal electrical bus was powered when the aircraft came to rest. If the aircraft had switched to the standby buses, only the pilot's Cl and copilot’s RMI should have displayed the correct compass headings. Other factors of particular significance tc substantiate that No, 2 engine was operating include: A. Vegetation was found throughout the sccondary gas path, in the primary gas path as far back as the diffuser case, and in the sixth and eighth-stage aircraft bleed system. Mud coating was found adhering completely through the engine from the inlet to the exhaust. B. All fuel nozzles were coated with dried mud except for the nozzle nut fuel exit holes. C. There was no evidence of overheat in the combustion area, turbine blades, or on the nozzle guide vanes. United Air Lines engine-out procedure is as follows: If an engine fails after reaching V4 speed, the takeoff will be continued. The climbout will be at V2 (if higher speed is already attained at the time of engine failure reduction to Vg speed is not necessary}}, with a 15° bank maximum and a inaximusn deck angle of 15°. On reaching 500 feer accelerate the aircraft to V2+15 knots and set flap position 1. At 190 knots set flaps 0. At a gross weight of $0,000 pounds operating from a field elevation of 14 feet and the aircraft in flight at an approximate height .. £50 feet and above Vo speed, a single-engine climb and a subsequent safe landing could kave been accomplished if che engine-out procedures had been followed.

[ANALYSIS] Such a high bending stress results in ratchet marks, indicating multiple fatigue origins, which increase in number with the degree of stress concentration and stress level. The number of cycles to failuce is primarily dependent on the level of this bending stress. In this case, the number of cycles is assumed to have been low. Inspection of the fuel pump drive shaft fracture surfaces revealed an extremely high number (at least 50) of fatigue origins. [t is not possible to determine the specific number of cycles from the initiation of the crack to failure since the stress level and time exposure arc not known, However, based on laboratory tests conducted at higher than normal bending load (180,060 p.s.i.}}, there is an indication that the failure could have occurred in less than 2,000 cycles from initiation of the crack, This was evidenced by the similarity between the rig failed shaft and engine failed shaft in thae the rig failure had at least 40 fatigue origins. These origins were very similar in appearance. This type of fracture could have occurred only at impact when the shaft was sulojected to an extreme misalignment over a short period of time. The location of the O-ring rub indicated thac the shaft was operating in an improperly aligned position, The roller bearing on the gearbox drive shafe nearest the fuel pump had pieces of the backside of the inner race broken out. The only plausible explanation is severe distortion of the gearbox housing which resulted in a thrust load from the outer race being teansferred through the rollers to the inner race. Microscepic analyses performed on the four first-stage nozzle guide vanes showed that the metal splattcr on the vanes was aluminum. Similar matevial was found lodged in the cooling air spigot of « vane and was deposited inside the vane core surfaces. The soutce of the aluminum was the chain link fence that the aircraft passed through, approximately 2,000 feet afser tauchdown. The melting point range of this type aluminum is approximately 1,150 to 1,250°F. in addition to the tactallurgica! analysis findings of the fucl pump drive shafe failure at impact, there are other factors to substantiate the fact that the No. 2 engine was operating at couchdown and throughout the overrun. The turbine inlet temperature decay graph shows at fucl shutoff the remperature will drop from 1,800°F. to 900°F. in 1 second and down to 600°F. in 2 seconds. It is recognized that turbine vane cooling rate is slower than the rate of decay of turbine intet gas temperature. However, the uniformity and degree of adherence of the aluminum splatter observed on the blades and vanes indicated that sutticient heat, pressure, rotation, and air flow were available upstream of the first-stage nozzle guide vanes co mele and fuse the alurninum splatter to the vanes and blades.

[ANALYSIS] It is recognized that turbine vane cooling rate is slower than the rate of decay of turbine intet gas temperature. However, the uniformity and degree of adherence of the aluminum splatter observed on the blades and vanes indicated that sutticient heat, pressure, rotation, and air flow were available upstream of the first-stage nozzle guide vanes co mele and fuse the alurninum splatter to the vanes and blades. This finding significantly demonstrates normal No. 2 engine operation after the aircrafe contacted the chain link fence. Additional evidence of No, 2 engine operation was an increase of engine noise level after touchdown aud audible on the cockpit voice recorder, (Appendix E.) The No, 2 engine thrust reverser was fully deployed and was subsequently functionally tested and found to operate properly A further indication that the engine and reverser were operating was the evidence that during the runway and off-runway roll the nose gcarwhcels tracked toward the right iain landing gearwheel track. (Appendix 1.) An engine oil pressure of 354 2 p.sig. is required to deploy the engine thrust reverser. Below this pressure, the reverser will not deploy. If the shaft had failed in flight, the engine oil pressure would have decreased below 37 p.s.i. in 4 seconds and below 35 p.s.i. in 5 seconds. The 13th-stage aircraft bleed modulating valve was closed. The 13th-stage bleed modulating valve would begin to open on decreasing power at the following conditions: N2 Percent of Vyas Percent Net Takeoff Knots pm. E.P.R. Thrust-lbs Net Thrust 0 85.4 1,392 6,660 49.0 102, 84.8 1.346 5,170 42.6 At fuel shutoff, a drop of Ng percent r.p.m. from 100 to 72 percent would occur in a maimum of 2 seconds. A drop from 1.95 EPR to 1.15 occurred in 1 second. A drop from 100 percent of takeoff net thrust to 25 percent would occur in 1 second. The 13th-stage bleed tnodulating valve closed position showed the engine was operating throughout the flight and during the reverse cycle until the aircraft came Co rest, Bocing Aircraft Company laboratory test showed that the generator would not carry an electrical load below 42.3 percent N3 speed. These values of No are for a deceleration of 180 r.p.m./sec. of Ny speed. At fucl shutoff, the percent of No dropped off to 41 percent in a maximum of 8 seconds. If the aircraft did have a double engine failure, the normal electrical generating systems would have been lost. Electrical power on the aircraft was available throughout the overrun. The only question is — was it coming from a normal bus or from a standby bur. and the airceafe battery.

[CONCLUSIONS] There was no evidence of factors which would have detracted from the pilot's physical ability to operate the aircraft. The aircraft was properly certificated and was maintained in accordance with approved maintenance procedures; however, inadequate maintenance was performed during replacement of the No. 5 cylinder of the left engine 4 days before the accident. The left engine turbocharger was malfunctioning before the accident. The turbocharger failed as a result of an imbalance and ceased operating after liftoff. A portior of the fractured exhaust valve guide from the No. 5 cylinder probably damaged the turbocharger's turbine wheel blades before the cylinder was replaced 4 days before the accident. The left engine sustained a substantial loss of power because the turbocharger ceased operating. The propeller was capable of normal operation. The right engine and propeller were capable of normal operation. The teft propeller, which was capable of normal operation, showed minimal evidence of rotation at impact and was transitioning to the feather position during the impact sequence. The right propeller was in the low pitch position and was producing power at impact. The aircraft's climb oerformance was very inarginal because of the high density altituce conditions and gross weight at takeoff. The pilot failed to feather the propeller in a timely manner and did net establish the aircraft in a minimum drag attitude following the power loss. The required sideslip technique used in engine-out procedures may not have been emphasized in the ptlot's flight training. 14... The horizontal and vertical peak crash loads were estimated to be within the limits of human tolerance. 15. There were no apparent seat or restraint system failures. 16. The fuel tanks ruptured during principal tree impact. 17. The fuel dispersed on impact and was probably ignited by hot engine components which caused an intense postimpact fire. 18. There were no significant traumatic injuries. 19. The cause of deeth of all occupants was attributed to burn injuries due to the intense posterash fire. 3.2 Prodabie Cause The National Transportation Safety Board determines that the probable cause of the accident was a substantial loss of power from the left engine at a critical point in the takeoff and the failure of the pilot to establish a minimum drag configuration which degraded the marginal single-engine climb performance of the alecraft. The loss of power resulted from seizure of the turbocharger fottowing progressive failure of the turbine wheel blades initiated by foreign object ingestion which had occurred previous to this flight and was not detected during maintenance on the engine 4 days before the accident. 4, RECOMMENDATIONS As a result of this investigation, the Safety Board reiterates Safety Recommendation A-79-80 which was issued to the FAA on May 17, 1978: Require that pilots Involved in 14 CFR 135 operations be thoroughly trained on the performance capabilities and handling qualities of aircraft when loaded to their maximum certificated gross weight or to the limits of their e.g. envelope, or both.

[ANALYSIS] The postimpact positions of the throttle, mixture, and propeller lever, as well as the propeller governor operating arm of the left engine, are evidence that, at sore point, he deliberately shut down the left engine and feathered its propeller. The two impact marks on the butt face of one propeller blade showed, however, that the left propeller blades were transitioning to the feather position at the time of the crash. Therefore, it is believed that the pilot waited until seconds before impact to feather the propeller. He also closed the emergency fuel shutoff valves before impact. Under the high density altitude conditions existing at the time of the accident and at an assumed airspeed of 105 KIAS, @ well-tuned engine with an dent Atte A El Ot A AE OT A OR A a RL OT ON rid rm ns sere Peers inoperative turbocharger would have produced about 155 shaft horsepower at 19 inches of manifold pressure and 2,135 rpm; the propeller blade angle would have been 17.7° Since the low pitch stop is set at 16.6°, the propeller blades would have been about 1° from the stop, and the constant speed features of the propeller would have caused the blades to move to the low pitch stop in the event cf a further loss of power or airspeed. The burning of oil and a rich fuel/air mixture in the left engine supports a conclusion that power was substantially reduced below the above values which resulted in more drag than thrust from the left propeller. Consequently, the aircraft's climb capability was degraded significantly because of the drag associated with the unfeathered propeller. In the event of engine roughtiess, the AFM directs the pilot to reduce “ power and analyze the condition before immediately feathering the propeller. A og reduction of power to idle in this case would have produced & maximum windmilling 4 propeller, and the aircraft's climb would have been degraded by 350 feet per minute, resulting in a minimum rate of descent of at least 190 feet per minute. Additionally, if the pilot used the sea level single-engine best rate of climb speed (V___) of 109 knots, mentioned in the AFM, he would have further degraded the el performance. The correct speed under the accident conditions was 99 knots. According to the AFM, at this speed, the aircraft would have climbed at 160 feet per minute, provided the aircraft was in a near zero sideslip attitude with the 4 propeller feathered. This rate, however, results in a less than 1° climb gradient. To achieve even a margiral climb capability, the pilot had to immediately feather the malfunctioning engine, establish a zero sideslip attitude, and fly at a V._. of 99 knots. Mismanagement of any one of these factors would have seriously Bffected his control of a barely manageable situation. Failure to feather the propeller under the conditions would have produced the greatest degradation in climb performance. An angle of bank of more than 5° in either direction would have also degraded the climb performance significantly.

[ANALYSIS] The former condition was the most probable occurrence of the imbalance condition. The airport manager's observation of a large umount of gray smoke emenating from the left engine 3 days before the accident is consistent with the damage in the turbocnarger. The turbine wheel imbalance would have disrupted its bearing seal and would have allowed passage of lubricating oil by the compresscr into the induction system to be burned in the combustion chambers of the engine. The witness' belief that the appearance of (he gray smoke was associated with a reduction in power is logical since there would be significantly lower combustion temperatures at idle power to burn the oil than at a high power setting. A seized turbocharger would have caused a significant drop in manifold pressure and power from the left engine. At full takeoff power, the manifold pressure developed is 40 inHg. With an inoperative turbocharger, the manifold pressure loss would be 3 to 5 inHg. below the normelly aspirated value of about 22 inHg. Also, there would be a corresponding imbalance in the fuel/air mixture ratio with the loss in manifold pressure due to a reduced induction airflow. The consequence of this condition would have been an overly rich mixture, a rough operating engine, an additional reduction in power, end the emission of black smoke. Evidenve of such a condition ‘existing in the left engine was observed on the exhaust valve heads, on the ceramic insulation of the spark plug electrodes, on spark gap surfaces, on components of the turbocharger, and from witnesses' reports, When such a condition occurs, it is necessary to manually lean the engine to achieve the best (uel/air mixture ratio. Although the AFM does not specifically direct such action in the event of engine roughness, it instructs the pilot to analyze the problem and secure the engine if roughness persists. The eircraft was also observed in a descending shallow left turn although the pilot had asked for a straightout departure. Impact occurred 45° to the left side of runway 21. The Safety Board concludes that the pilot was confronted with a partial but substantial loss of power from the left engine shortly after liftoff due to the failure of the turbocharger. He apparently was able to initially establish a climb; however, the aircraft reportely did not climb more than 200 feet above the runway. The postimpact positions of the fuel selector valve handles indicate that he may have suspected a fuel problem. If this was the case, he would only have had to crossfeed the left engine from the tight tank. The postimpact positions of the throttle, mixture, and propeller lever, as well as the propeller governor operating arm of the left engine, are evidence that, at sore point, he deliberately shut down the left engine and feathered its propeller. The two impact marks on the butt face of one propeller blade showed, however, that the left propeller blades were transitioning to the feather position at the time of the crash.

[ANALYSIS] Metallurgical examination of the left engine turbocharger disclosed that it had seized suddenly but that the engine had continued to operate for some time following failure. There were hot gas erosion holes in one-half of the turbine wheel shroud and silhouette imprints of the turbine wheel blades on the housing. Both conditions indicate thet hot exhaust gases impinged for an appreciable time on the turbine wheel while it was stationary. Further, metal transfer between the compressor impeller blades and housing and the turbine wheel blades and housing indicates that a severe imbaiance of the turbine wheel and compressor existed within the turbocnarger assembly. The wear on the center bearing and the deformation of the turbine wheel shaft, adjacent to the shaft fracture, are also consistent with an imbalance condition. Only a sniall portion of the fracture on the turbine shaft was clean or free of deposits; the remainder contained heavy, lead rich exhaust gas residues. The deposits indicated that the majority of the shaft fracture had been exposed ‘o exhaust gases for some time. The discolored fracture surface exhibited cracking features which had originated at the bottom of the piston ring groove with fatigue crack propagation toward the irside diameter of the shaft. However, the crescents indicative of this type of cracking did not extend to the inside diameter. Thus, the final fracture probably occurred rapidly because of an overstress condition. As a result, the fracture was subjected to hot gases escaping through the turbine whee: shroud. The material found on the outside diameter of the broken shaft, adjacent to the turbine wheel, and the niston ring groove sidewall had a chemical composition consistent with the piston ring. This indicated that the ring was smeared between the shaft and the inside diameter of the center housing, This condition could have occurred from unbalanced rotational forces. The Safety Board was unable to determine precisely the source that initially produzed the imbalanced condition in the turbocharger assembly. uwevert, since all the turbine wheel blades were broken near the tips, the blades probably were broken as a result of foreign object damage (FOD). Such damage would produce severe turbine wheel imbalance and the conditions found in the turbocharger assembly, Similar conditions have been documented by the manufacturer in known FOD damaged turbochargers. While it is possible tha’ the arate yhlemyagehahmneemae: AR RENIN ETE aie bo, ~-2i- turbine shaft may have fractured totally, or in part by itseif, precipitating the imbalance, the Safety Board concludes that, because of the exhaust valve guide findings, the weight of evidence indicates that the lower portion of the No, 5 cylinder exhaust valve guide caused damage to the turbine blades which produced an imbalanced condition and led to seizure of the turbocharger. The former condition was the most probable occurrence of the imbalance condition. The airport manager's observation of a large umount of gray smoke emenating from the left engine 3 days before the accident is consistent with the damage in the turbocnarger.

[ANALYSIS] Mismanagement of any one of these factors would have seriously Bffected his control of a barely manageable situation. Failure to feather the propeller under the conditions would have produced the greatest degradation in climb performance. An angle of bank of more than 5° in either direction would have also degraded the climb performance significantly. Since only a small reduction in power below a normally aspirated engine condition would have caused more drag than thrust, the pilot may have observed what he believed to be sufficient power indications on the manifold and rpm gages for him to attempt to og regain full power from the engine. However, under the circumstances, the aad propeller would have been producing more drag than thrist, which probably we- not E fully appreciated by the pilot. It is also believed that the importance of establishing the aircraft in the required sideslip attitude may not have been stressed during the pilot's training. The evicence suggests that the pilot did not feather the propeller in time and did not exercise the precise control necessary to obtain the available performance of the aircraft, probably because of distractions associated with his attempts to regain full power from the engine. Therefore, considering the significant climb degradations of a partially windmilling propeller and drag due to sideslip, a failure to feather the propeller and establish the necessary sideslip attitude in a timely manner made this accident inevitable. Notwithstanding the above circumstances, the Safety Board believes that the pilot's ability to sustain flight was marginal and, therefore, had he been more timely in his actions, it is not certain that an accident would have been avoided. In view of the very marginal single-engine climb performance available under the conditions of the accident and the precise flying necessary to achieve this performance, the Safety Board, in summary, reiterates its conclusions previously reported in a special study 2/--that at the present time light twinengine safety during critical engine failure situations depends almost exclusively upon pilot proriciency. The pilot's ability to immediately assess the situation and react instinctively in a correct manner is the level of safety factor upon which the public is dependent in most part 135 operations. In this regard, pilot training cannot be overemphasized. This accident and another which occurred on March 1, 1979, 3/ underscore the need for operators to carefully evaluate the proficiency of all pilots in their employ who nave not had previous training and experience in twin-engine propeller driven aircraft. This fact alone, considering his previous military experience, might explain why the pilot did not instinctively react in the proper manner. The Safety Board also recognizes the criticality of single-engine training in these types of aircraft and, believes that the greatest potential for prevention of these types of accidents ‘Is in the area cf increased’ single-engine performance of light twin-engine aircraft. The available evidence indicates that the structural crashworthiness of the aircraft was not compromised during the impact sequence and that there remained adequate occupiable space for all occupants. The seats and restraint systems appeared to have remained intact and should have provided restraint for all occupants. --- Footnotes: [3/ 4-inch sideplay and that the exhaust valve guide was found fractured. Although the niaintenance personnel stated that the lower portion cf the fractured valve guide was not removed from the cylinder, they were not able to confirm that it remained in the cylinder head. Because of the severe sideplay in the exhaust valve, the fracture of the valve guide, and the damaged cylinder boss area which surrounded the exhaust valve, the Safety Board concludes that the lower pvstion of the guide was missing before the No, 5 cylinder was removed from the engine. Moreover, we believe that the cause of the sideplay was obvious and, if properly investigated, would have disclosed that a portion of the guide was either broken or missing under circumstances which could have caused further damage to the]

[CONCLUSIONS > FINDINGS] All the fractured surfaces were consistent with overstress separations. 4. Several pieces of foreign material, made of M~50 alivy steel, were found in the engine and engine debris. The pieces were battered severely. Three of these nieces were found in the low pressure turbine's stage 2 rotor cavity; and two of these pieces were found in the engine's stage 1 toroid cavity. , ( {{ f i 3 £ 74 t 6. 11. 12. 13. 14. -24~ The stage 1 to stage 2 rotor bolt heads were smeared with this M-50 alloy steel. The failure of the stage 1 to stage 2 rotor bolts and the subsequent failure of the stage 1 Jow pressure turbine rotor was caused by the impingement of the pieces of M-50 alloy steel un the rotor bolts. The engine's main bearings are inade of M-50 alloy steel. All of the main engine bearings forward of the low pressure turbine section were undamaged, The part of the low pressure turbine cavity where the M-50 fragments were found is sealed from the path of gas flow through the ngine. The M-50 fragments could not have migrated either forward or aft into the low pressure turbine cavity while the engine wus in operation, The M-50 fragments found inside the low pressure turbine rotor cavities were not part of this engine. Debris from the stage 1 low pressure turbine rotor disk eut the slat follow-up cable which connects the inboard slni actuator mechanism to the servo valve. The right wing outboard slat group was almost completely retracted as a result of this damage. The Nos. 1 and 3 hydraulic systems were inoperative from lack of fluid. The firewall fuel shut off valve actuating cables and the fire extinguishing discharge line from the two agent bottles for the No. 3 engine were severed. The No. 3 engine's thrust lever control cable was severed. All three generator AC buses and all TR buses remained powered. The stall warning systems and flight director systems remained operative. V2 speed was 6 KIAS faster than the slats retracted stall speed. Had the engine failure occurred at or after V1 speed, the flighterew should have been able to continue the takeoff safely. The DC-10-30 cngine installation configuration and its leading edge wing slat system were certified in accordance with the applicable regulations and Special Conditions. The present certification regulations do not establish precise standards and guidelines for analyzing the effects of an uncontained rotor burst. 3.2 Probable Cause The National Transportation Safety Board uecermines that the probable cause of this accident was the failure of quality control inspections to detect the presence of foreign material in the low pressure turbine cavity during the reassemb’y of the low pressure turbine module after installation of the stage 1 low pressure turbine rotor disk.

[ANALYSIS] This cross sectional area would "neck down" during a turbine disk overspeed. In addition, the metallurgical examination of the pieces of the failed disk showed no evidence of preexisting fatigue or cracking and that the fracture surfaces were consistent with overstress separations. In order for the stage 1 low pressure turbine rotor to have accelerated to the overspeed condition, it would had to have been free of the torque load of its rotor assembly. This freedom was accomplished by the failure of some of the 60 bolts which attached the stage 1 rotor to the stage 2 rotor at the bolted flanges of their spacer arms. Five pieces of foreign material--M-50 alloy steel-- were found in the engine. Three of these pieces were found in the low pressure turbine's stage 2 rotor cavity; two pieces were found in the engine's stage 1 toroid cavity. In addition, the stage 1 to stage 2 bolt heads were smeared with M-50 alloy steel. This condition indicated that, over a relatively lcag period, the impingement of this material damaged a sufficient number of these 60 bolts to destroy the bolts' capability to hold the stage 1 rotor to the stage 2 rotor. The bolts failed, the rotors separated, and the stagei rotor was free of its rotor assembly load. Asa result, the stage 1 rotor disk oversped during the high takeoff power demand on the engine. The overspeed resulted in tne fragmentation of the rotor disk and exit of the fragments through the low pressure turbine casing. Consequently, the Safety Board concludes that the faiture of the stage 1 low pressure turbine rotor disk was caused by the foreign material in the engine. The five frrngments of M-50 alloy steel found inside the low pressure turbine module were severely oxidized and battered indicating that they had been exposed to heat and impact damage for a considerable period. The main engine bearings of the CF6 engine fleet are made with M~50 alloy steel; however, all the main bearings forward of the low pressure turbine section were undamaged and all the engine bearings’ inner races were intact. In addition, the part of the low pressure turbine section where these fragments were found is sealed from the path of gas flow through the engine. Consequently, this material could not have migrated either aft or forward into the low pressure turbine cavity while the engine was in operation. The evidence appears conclusive that either these fragments were inside the low pressure turbine module when it was installed on the engine or they got inside the turbine module the last time it was disassembled and were not detected during maintenance. Further, the battered condition of these fragments precluded all attempts to match them with the toois used in the engine build-up procedure, or with any other type M-50 alloy steel tool, consequently the Safety Board could not identify the source of these M-50 alloy stee) fragments, Exainination of the damaged low pressure turbine indicated that the initial exit of the rotor disk fragments occurred just outboard of the top of the engine.

[PROBABLE CAUSE] The present certification regulations do not establish precise standards and guidelines for analyzing the effects of an uncontained rotor burst. 3.2 Probable Cause The National Transportation Safety Board uecermines that the probable cause of this accident was the failure of quality control inspections to detect the presence of foreign material in the low pressure turbine cavity during the reassemb’y of the low pressure turbine module after installation of the stage 1 low pressure turbine rotor disk. The foreign material in the low pressure turbine cavity damaged the bolts holding the stage 1 low pressure turbine rotor disk and stage 2 low pressure turbine rotor disk together. The bolts failed at high engine thrust and the stage 1 low pressure turbine disk separated from the low pressure turbine rotor assembly, oversped, and burst. we , ' - ' we + —. a to a . —_ fe ~ _ a ad - aie - a Fa : . r . ra Lo . tere ee Gp care i tai me OP eae . - ee nee Ts teenene me pumas: G. A. PATRICK BURSLEY, Mernber, did not participate, April 6, 1982 ~25—- 4. RECOMMENDATIONS ar nrenine Rimeatapneit Tea As a result of its investigation of this accident, the National Transportation Safety Board recommended that the Federal Aviation Administration: Expedite the publication of guidance material for acceptable means of compliance with 14 CFR 25.903(d)(1), which includes compliance documentation by failure mode and effect analysis, provides for rotor fragment energy levels and paths based on cases of severe in-service damage, and reflects advances in analytical techniques and concepts which have taken place since certification prograins of the early 1970's. (Class ll, Priority Action) (A-82-38) Actively encourage research and development in containment technology and engine reliability, including basic design concepts, manufacturing processes, and maintenance factors to detect and prevent impending failures. (Class II, Priority Action) (A-82- 39) BY THE NATIONAL TRANSPORTATION SAFETY BOARD /s/ JAMES E. BURNETT, JR. Chairman /s/ FRANCIS H. McADAMS Member /s/ PATRICIA A, GOLDMAN Member eee ~ 7 et wee . eT een OE - a “ . . , aa . * 2 oF I \ ' . I a iF a ns seas WAG axiy RHO See Che aM eC a am pS mem bin behinds 5S. APPENDIXES APPENDIX A INVESTIGATION AND HEARING 1. Investigation The National Transportation Safety Board was notified of the incident about 1740 e.d.t. on september 2%, 1981, and immediately dispatched a partial investigative team to the scene.

[ANALYSIS] Consideration of the effect of an uncontained rotor failure is now provided for iti L4 CFR 25.571(a) which states, in part, "an evaluation of the strength, detail design, and fabrication must show that catastrophic failure due to....accidental damage will be avoided throughout the operational life of the airplane. This evaluation must be conducted in accordance with the provisions of paragraphs (b) and (e) of this section." 14 CFR 25.571(b) requires the applicant to determine the provable location and modes of damage which could result from accidental damage and 14 CFR 25.571{{e) states, in part, "The airplane must be capable of successfully completing a flight during which likely structural damage occurs as a result of,... (3) uncontainec engine failure." The only published guidance material for showing complianee with this regulation is contained in Advisory Circular (AC) 25.571-1, Damage Tolerance and Fatigue Evaluation of Structure. Paragraph 4(p)(2) of the AC states, in part, "cin the case of uncontained engine failures, the fragments and paths to be considered should be consistent with those in showing compliance with 14 CFR 25.903(d)(1) of the FAR's, and with typical dainage experienced in service." Thus, the regulations now require an aireraft manufacturer to consider the possibility of an uncontained rotor failure and to minimize the effeet the rotor fragments will have on the capability of the aircraft to continuc safe flight, Based on information nrovided by the applicant, the FAA thei determines whether the standards contained in (4 : FR 25 have been satisfied. Although the regulations do not establish uniform guidelines or standards for determining roto. fraginent size, path, cnergy, or a method of documenting these rotor burst analyses, FAA Order 8110.11, issued on November 19, 1975, provided information on design considerations for minimizing uneontained rotor failure damage. The information was for use in regional flight standards offices and the Aircraft Engineering Division in the Western Region, However, the information did not include analytical methods or probability calculations. Although engine rotor failures are relatively rare, their potential for causing serious damage to the aircraft or its systems is quite high. With regard to high bypass ratio engines, there have been 23 turbine ar compressor rotor disk failures during the past 12 years, many of which have damaged ihe airernft or its systems substantially. 7/ Society of Automotive Engineers Acrospace Information Report AIR 1537 Report. on Aircraft Engine Containment 1977. Tue rate of these failures have been relatively constant, and moreover, current materials and fabrications technology is such that containment of rotor disk fragments is not foreseeable in the near future. Given the technological limitations on containment, the Safety Board has no reason to believe that these failures will cease in the future, either on the current high bypass ratio engines or their successors. Therefore, the Safety Board is concerned that adequate aircraft design precautions are taken to minimize the hazards associated with uncontained rotor disk fragments. --- Footnotes: [7/ Society of Automotive Engineers Acrospace Information Report AIR 1537 Report. on Aircraft Engine Containment 1977.]

[ANALYSIS] Further, the battered condition of these fragments precluded all attempts to match them with the toois used in the engine build-up procedure, or with any other type M-50 alloy steel tool, consequently the Safety Board could not identify the source of these M-50 alloy stee) fragments, Exainination of the damaged low pressure turbine indicated that the initial exit of the rotor disk fragments occurred just outboard of the top of the engine. Thereafter, the fragments left the engine aroind the circumference of the engine and produced the air frarne damage described previcusly. 2.2 Slat Retraction and Takeoff Performance The No. 1 and No. 3 hydraulic reservoirs were empty and the No. 1 hydraulic system's slat retract line was separated. The No. 3 systetns pump suction line was separated and the system was slowly depleted by reservoir “head pressure." The right outboard slats were found in an intermediate position but almost fully retracted. This wrobably oeeurred for the following reason:: 1. The rotor fragments cut the slat follow-up cable. This cable drives the slat servo valve to the closed position when the slat reaches the position commanded by the pilot. When the cable was cut the slat servo valve was then free to go to any wosition and, in this ease, it traveled to a position wherein the hydraulic fluid from systems Nos. 1 and 3 was ported to the retract side of the slat actuator cylinder. 2. The hydraulic fluid in system No. 1 was pumped overboard at normal system pressure. 3. The hydraulic fluid in system No. 3 was depleted by reservoir "head pressure." Since the engine failure and slat retraction did not occur during the critical portion of flight between V1 and V2 speeds, speculation on what might have occurred had the takeoff been continued does net constitute any part of the cause and effect of the aecident. However, given the background of slat problems in this area, the Safety Board believes that soine of the facts relating to a continued takeoff should be discussed. . Dey ’ ‘ ae a to Vo ' ! “s « \ ¢ i i t . i ha i on hob: ng + : fh + -21i- The No. 2 hydraulic system was intact; therefore, the aircraft was capable of continued flight although with some flight controls at a reduced level of effectivenass. All three generator buses were operative. The evidence was conclusive that the right wing's outboard slat sensor was providing aecurate position information to the DFDR; therefore, this saine information would have been provided to the ATSC's and the Slat disagreement warning light system. Consequently, the Safety Board concludes that the slat disagreement warning litht operated properly and that the flighterew did not notice that this light was iNuminated during the rejecied takeoff because they were preoccupic:d with higher priority crew tasks.

[ANALYSIS] The engine manufacturers are governed by the provisions of 14 CFR 33; however, 14 CFR 33 only requires that the compressor and rotor cases provide "...for the containment of damage from rotor blade failure;' it does not levy any requirement to contain a rotor disk fragment. Part 33 addresses the latter problem by requiring the engine manufacturers to guard against rotor disk burst by establishing service life limitations for the engine rotors, and to demor strate that the rotors can sustain structural integrity up to, and including, established overspeed :nd overtemperature limitations. Howeve., udustry-wide studies of all turbine engine experience have indicated consistently that the rotor failure problem, while not statistically alarming (.66 failures ~22-- per million engine hours during 1962-1975 and a factor in 0.22 percent of all fatalities 7/), had the potential for causing serious aircraft damage. Part 25 attempts to address this problem by requiring the manufacturer to take design precautions to minimize damage which coulda occur following an uncontained engine rotor burst. In addition, 14 C¥R 25 also has been amended and ampiified since 1970 and now reflects the requirements which were contained in Special Condition No. 25-18-WE 7(L). 14 CER 25.903(d)(1) now states, in part, "Design precautions must be taken to minimize the hazards to the aircraft in the event of an engine rotor failure....". However, the regulation is general and there is no published guidance material setting forth acceptable methods for demonstrating compliance with the requirement. Currently, each applicant must show the design precautions which have been taken to minimize the hazards of rotor failure on his aircraft, end a determination of acceptance is based on a subjective evaluation of the adequacy of the analysis and the effectiveness of the designed precautions. Consideration of the effect of an uncontained rotor failure is now provided for iti L4 CFR 25.571(a) which states, in part, "an evaluation of the strength, detail design, and fabrication must show that catastrophic failure due to....accidental damage will be avoided throughout the operational life of the airplane. --- Footnotes: [7/ Society of Automotive Engineers Acrospace Information Report AIR 1537 Report. on Aircraft Engine Containment 1977.]

[ANALYSIS] 2. ANALYSIS 2.1 General The flightcrew was trained, certificated and qualified to conduct the flight, and the flight was conducted in accordance with applicable FARs and company requirements. The flight attendant also was appropriately trained and qualified. The flightcrew was in good health and held the proper FAA medical certificates. There was no evidence that the performance of any crew member was impaired by alcohol, drugs, or fatigue. The airplane was maintained in accordance with applicable FARs and company Operations Specifications. A review of the airplane’s maintenance records and operating history did not reveal any maintenance discrepancies or mechanical anomaly that would have either caused or contributed to the accident. Evidence from the CVR, FDR, and examination of the powerplants, reduction gear boxes, and propellers indicated that the engines were operating normally during the flight until the loss of a major portion of one propeller blade on the left engine. After the propeller blade separation, the combination of the resulting loss of left engine thrust, increased drag from a deformed engine nacelle and the three blades retained in the propeller hub, and added frontal drag from external sheet metal damage, degraded airplane performance, preventing the flightcrew from arresting the airplane’s descent or making rapid changes in its direction of flight making a forced landing necessary. The Safety Board concludes that because of the severely degraded aircraft performance, the flightcrew’s actions were reasonable and appropriate during their attempts to control and maneuver the airplane throughout the accident sequence and were not a factor in this accident. 2.2 Analysis of the Propeller Blade Failure The Safety Board concludes that one of the four blades from the left engine propeller separated in flight because a fatigue crack that originated from multiple corrosion pits in the taper bore surface of the blade spar propagated toward the outside of the blade, around both sides of the taper bore, then reached critical size. (See Section 1.16.1.) 54 Results of investigations conducted in two previous propeller blade failures in 1994, one in Brazil with this model blade and the other in Canada with a similar model blade, indicated that corrosion was produced when entrapped moisture reacted with residual chlorine in a bleached cork used to retain the lead wool in the taper bore hole of the propeller. The accident blade exhibited a nearly continuous layer of oxide deposits on the initial 0.049 inch of the crack depth. These deposits contained a substantial amount of chlorine. The Safety Board found that the ASA propeller blade contained corrosion damage (pitting) in the taper bore and the oxide layer in the origin area of the fatigue crack in the separated ASA propeller blade, as did the two previous failed propellers.

[CONCLUSIONS > FINDINGS] 3.1 Findings 1. The flightcrew was trained, certificated and qualified to conduct the flight, and the flight was conducted in accordance with applicable Federal Aviation Regulations and company requirements. 2. The flightcrew was in good health and held the proper FAA medical certificates. There was no evidence that the performance of any crewmember was impaired by alcohol, drugs, or fatigue. 3. ASA maintained the airplane in accordance with applicable Federal Aviation Regulations and company Operations Specifications. 4. After the propeller blade separation, the combination of the resulting loss of left engine thrust, increased drag from a deformed engine nacelle and the three blades retained in the propeller hub and added frontal drag from external sheet metal damage degraded airplane performance preventing the flightcrew from arresting the airplane’s descent or making rapid changes in its direction of flight making a forced landing necessary. 5. One of the four blades from the left engine propeller separated in flight because a fatigue crack that originated from multiple corrosion pits in the taper bore surface of the blade spar propagated toward the outside of the blade, around both sides of the taper bore, then reached critical size. 6. Because of the severely degraded aircraft performance following the propeller blade separation, the flightcrew’s actions were reasonable and appropriate during their attempts to control and maneuver the airplane throughout the accident sequence, and they were not a factor in this accident. 76 7. Hamilton Standard’s engineering decision to use the PS960A blending repair to remove ultrasonic indications caused by a shotpeened taper bore surface was technically reasonable. 8. The manner in which the unapproved extension of PS960A was documented and communicated within Hamilton Standard, and the lack of training on the extension, created confusion and led to misapplication of the blending repair to unshotpeened blades with unexplained ultrasonic indications, allowing the accident blade to be placed back into service with an existing crack. 9. The sanding marks left by the PS960A blending repair did not contribute to the initiation of the fatigue crack in the accident blade. 10. The failure to restore the taper bore surface to the original surface finish, as required by PS960A, was a factor that caused the reduction of the ultrasonic indication that allowed the blade to pass the final ultrasonic inspection and to be returned to service. 11. The borescope inspection procedure developed and used by Hamilton Standard in June 1994 to inspect returned blades that had rejectable ultrasonic indications for evidence of cracks, pits, and corrosion was inadequate and ineffective. 12.

[ANALYSIS] Although it is not explained in AC 20-66, testing a propeller blade at its cutoff repair limit is conducted presumably to ensure that the decrease in blade length does not alter the natural frequency to the extent that a resonant vibration condition could be entered while operating within the normal rpm range. AC 20-66 considers the expected loss of mass of a metal propeller blade following blend repair; however, the AC does not consider the expected mass gain with age of a composite propeller blade. Composite blades may gain mass with age with the addition of layers of paint, introduction of moisture, and patch repairs. A review of AC 20-66, which includes a detailed discussion of the propeller vibratory phenomenon, does not explain that a propeller blade’s natural 66 vibratory response varies with varying conditions (i.e. mass gain, mass loss, variations in airfoil shape, etc.) and that adequate margin from a potentially coincident excitation frequency should be maintained. Consequently, the Safety Board concludes that the AC does not provide guidelines for adequate margin between a propeller blade’s natural frequencies and its potentially coincident excitation frequencies over the life of the blade. Therefore, the Safety Board believes that the FAA should revise AC 20-66 to include the vibratory testing of composite propeller blades that have been previously operated for a substantial number of service hours, and composite blades that have been altered to the limits set forth in FAA-approved repair manuals to determine the expected effects of age on propeller vibration and provide guidelines for rpm margin between a propeller blade’s natural frequencies and the excitation frequencies associated with propeller operation. 2.5 Effect of Blade Failure and Analysis of Terminating Action The forward half of the fractured front inlet case, found along the wreckage path, was attached to the reduction gear box (RGB) of the left engine. The RGB case was also fractured, empty of oil, and some internal gearbox components were missing. The in-flight fracture of the RGB case most likely resulted in the loss of the gearbox components and the venting liquid reportedly seen by the passengers. Damage to the forward engine mounts and the front frame of the left engine nacelle indicated that the propeller RGB first separated upward at the inboard mount location, followed by the aft and outboard failure at the outboard mount location. Failure of the engine mounts were the result of overload stresses. The imbalance loads associated with the separated blade on the accident airplane apparently exceeded the design strength of the RGB attachment points. In this instance, it appears that the RGB was retained on the airplane in an outboard displaced position, as observed by the passengers, from the time of propeller blade separation until the beginning of the initial ground impact. Embraer’s postcertification (September 1984) testing determined that the nacelle would not withstand a mid-blade or full-blade segment loss. To date, there have been four blade separations--three from fatigue cracks that initiated in the taper bore.

[CONCLUSIONS] The airplane continued its descent and was destroyed by ground impact forces and postcrash fire. The captain and four passengers sustained fatal injuries. Three other passengers died of injuries in the following 30 days. The first officer, the flight attendant, and 11 passengers sustained serious injuries, and the remaining 8 passengers sustained minor injuries. The National Transportation Safety Board determines that the probable cause of this accident was the in-flight fatigue fracture and separation of a propeller blade resulting in distortion of the left engine nacelle, causing excessive drag, loss of wing lift, and reduced directional control of the airplane. The fracture was caused by a fatigue crack from multiple corrosion pits that were not discovered by Hamilton Standard because of inadequate and ineffective corporate inspection and repair techniques, training, documentation, and communications. Contributing to the accident was Hamilton Standard’s and the Federal Aviation Administration’s failure to require recurrent on-wing ultrasonic inspections for the affected propellers. Contributing to the severity of the accident was the overcast cloud ceiling at the accident site. vi Safety issues in the report focused on manufacturer engineering practices, propeller blade maintenance repair, propeller testing and inspection procedures, the relaying of emergency information by air traffic controllers, crew resource management training, and the design of crash axes carried in aircraft. Recommendations concerning these issues were made to the Federal Aviation Administration. NATIONAL TRANSPORTATION SAFETY BOARD WASHINGTON, D.C. 20594 AIRCRAFT ACCIDENT REPORT IN-FLIGHT LOSS OF PROPELLER BLADE FORCED LANDING AND COLLISION WITH TERRAIN ATLANTIC SOUTHEAST AIRLINES, INC., FLIGHT 529 EMBRAER EMB-120RT, N256AS, CARROLLTON, GEORGIA AUGUST 21, 1995 1. FACTUAL INFORMATION 1.1 History of Flight On August 21, 1995, about 12531 eastern daylight time, an Empresa Brasileira de Aeronautica S.A. (Embraer) EMB-120RT, N256AS, airplane operated by Atlantic Southeast Airlines Inc., (ASA2) as ASE3 flight 529,4 experienced the loss of a propeller blade from the left engine propeller while climbing through 18,100 feet. The airplane then crashed during an emergency landing near Carrollton, Georgia, about 31 minutes after departing the Atlanta Hartsfield International Airport (ATL), Atlanta, Georgia. The flight was a scheduled passenger flight from ATL to Gulfport, Mississippi (GPT), carrying 26 passengers and a crew of 3, operating according to instrument flight rules (IFR), under the provisions of Title 14 Code of Federal Regulations (CFR) Part 135. The flightcrew declared an emergency and initially attempted to return to Atlanta. The flightcrew then advised air traffic control (ATC) that they were unable to maintain altitude and were vectored toward the West Georgia Regional Airport (CTJ), Carrollton, Georgia, for an emergency landing.

[PROBABLE CAUSE] This accident illustrates that critical information regarding time available to prepare the aircraft for an emergency landing or impact is not being considered and communicated among flight and cabin crewmembers. 23. There should be standards governing the design of crash axes required to be carried aboard passenger-carrying aircraft. 79 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the in-flight fatigue fracture and separation of a propeller blade resulting in distortion of the left engine nacelle, causing excessive drag, loss of wing lift, and reduced directional control of the airplane. The fracture was caused by a fatigue crack from multiple corrosion pits that were not discovered by Hamilton Standard because of inadequate and ineffective corporate inspection and repair techniques, training, documentation, and communications. Contributing to the accident was Hamilton Standard’s and FAA’s failure to require recurrent on-wing ultrasonic inspections for the affected propellers. Contributing to the severity of the accident was the overcast cloud ceiling at the accident site. 80 4. RECOMMENDATIONS As a result of the investigation of this accident, the National Transportation Safety Board makes the following recommendations: --to the Federal Aviation Administration: Require Hamilton Standard to review and evaluate the adequacy of its tools, training, and procedures for performing propeller blend repairs, and ensure that those blend repairs are being performed properly. (A-96-142) Review the need to require inspection (“buy back”) after the completion of work that is performed by uncertificated mechanics at Part 145 repair stations to ensure the satisfactory completion of the assigned tasks. (A-96-143) Revise Advisory Circular 20-66 to include the vibratory testing of composite propeller blades that have been previously operated for a substantial number of service hours, and composite blades that have been altered to the limits set forth in FAA-approved repair manuals to determine the expected effects of age on propeller vibration and provide guidelines for rpm margin between a propeller blade’s natural frequencies and the excitation frequencies associated with propeller operation. (A-96-144) Require that Hamilton Standard consider long-term, atmosphericinduced corrosion effects and amend the Component Maintenance Manual (CMM) inspection procedure to reflect an appropriate interval that will detect any corrosion within the taper bore. (A96-145) Require Hamilton Standard to review and, if necessary, revise its policies and procedures regarding 1) internal communication and documentation of engineering decisions, and 2) involvement of the Designated Engineering Representative (DER) and FAA, and to ensure that there is proper communication, both internally and with the FAA, regarding all significant engineering decisions. (A96-146) 81 Include an article in the Air Traffic Bulletin and provide a mandatory formal briefing to all air traffic controllers regarding the necessity and importance of notifying crash, fire and rescue personnel upon a pilot’s request for emergency assistance.

[ANALYSIS] Embraer’s postcertification (September 1984) testing determined that the nacelle would not withstand a mid-blade or full-blade segment loss. To date, there have been four blade separations--three from fatigue cracks that initiated in the taper bore. The first blade separation (Inter-Canadien) resulted in RGB and 67 propeller separation, and the assembly fell to earth. During the second blade separation (Nordeste), the RGB and propeller assembly remained in place. During the third separation (the Luxair accident), in which a fracture occurred in the blade shank area, the RGB and propeller assembly again fell from the airplane. During the fourth blade failure (this accident), the RGB rotated out of position, and resulted in degraded aerodynamic performance and a fatal accident. Although in two of the occurrences, the RGB and propeller fell clear and did not seriously compromise the airplane or degrade its performance, all of the occurrences clearly placed the airplane and its occupants at potential serious risk. On four occasions, stresses on blades with flaws (corrosion pits or mechanical damage) have produced a blade separation even though the propeller was certificated based on the assumption of an unlimited life. Because the current regulations do not require that an airframe survive if a blade breaks, and because Embraer has determined that the EMB-120 cannot survive the loss of a mid-blade or full-blade segment, minimizing the possibility of a propeller blade separation is imperative. To prevent future failures, it is essential that stress risers in the form of corrosion or mechanical damage are not permitted to occur on any propeller blade. The Safety Board concurs that the taper bore repair procedure specified in the March 1996 Service Bulletins (and required by ADs) should have restored the surface of the taper bore of all existing propellers to a nearly new condition. Also, because Hamilton Standard has prohibited the use of the mechanical lead-removal tools during routine blade balancing, the likelihood of future inadvertent mechanical damage has been greatly reduced. However, while the terminating taper bore repair procedure should detect and eliminate any chlorine-induced corrosion or mechanical damage, the Safety Board is concerned that exposure to small amounts of moisture or other atmospheric elements during routine maintenance, the recurring inspection procedure set forth in the CMM, periods of low utilization, or long-term storage may allow atmospheric-induced corrosion to begin in the taper bore. The Safety Board is aware of reports of corrosion and cracking in the taper bores of P-3 and C-130 propellers associated with long-term storage. Because of this, the Safety Board concludes that despite all the actions taken by Hamilton Standard and the FAA to date, there is a continuing potential for corrosion to develop in taper bores of the affected Hamilton Standard propeller blades.

[ANALYSIS] Flight characteristics can alter appreciably the angle of air inflow to the propeller. Another factor which could affect propeller vibratory stresses is propeller proximity to the fuselage. No dynamic testing relative to this aspect had been conducted on the DH~-114. 2. ANALYSIS AND CONCLUSIONS Analysis The crewmembers were qualified and certificated in accordance with existing regulations. The airplane was maintained in accordance with an FAA-approved aircraft inspection prograa. The Safety Board concludes from the physical evidence that the No. 1 propeller of the No. 2 engine separated as a result of multiple fetigue cracks around the propeller blade shank. These cracks indicate that re blade had been subjected to abnormal vibrations over a prolonged period. The hardness and microstructure of the metal was as specified. There was no evidence of mechanical damage or of a material defect in the area in which tne fracture originated. Previous failures of this type of blade had been attributed to pressure caused by excessive contact between the shank clamp and the shank fillet; however, the Safety Board found no evidence of such »ressure on the hub surface. The surfaces that could have contacted the clamp were not damaged. The Safety Board could not determine the exact cause of the abnormal vibrations. However, the vibrat’.ons which are transmitted to the propeller by excessively worn counterweight pins and bushings in the engine's crankshaft can be supported by the past experience. Although the Safety Board could not determine the past history ot the failed blade, it is possible that the blade had been installed on an engine which had counterweight pins and bushings that were worn beyond acceptable limits and that excessive stresses weve imposed and caused a crack nucleus in the propeller shank. If this was the case, the crack nucleus should have been found when the propeller was last overhauled. However, based on its evaluation of the propeller overhaul facility, the Board believes that it is probable that facility personnel would have failed to detect the defect during overhaul, The carrier had been advised as early as 1970 of the manufacturer's recommendations on operating times between overhauls. The Safety Boara has found that {{t is not unusual for the carrter and local FAA authorities to adjust such recommendations to suit both service experience and economic cons iderat/ons. Since the FAA was aware of the potential hazard to fligirt which cculd result from noncompliance with the manufacturer's recommendations of 1976, it was incumbent upon them to make this clear to the operator and rake such recommendations mandatory. The Board believes that the manufacturers of the propeller aid the engine as well as the respective FAA Region share the responsiblity for insuring compliance with mandatory inspection and maintenance procedures, which, to their knowledge, can adversely affect safe flight.

[(a) FINDINGS] Tne No, 1b blade of the No. 2 propeller separated at the blade shank curing the takeoff rol. Vibratory stresses caused a fatigue fracture in the biade's shank area, The crack existed before Flight 303 began the takeoff roll but was not detectable during preflight inspection. The manufacturer of the propeller and the manufacturer of the engine were aware of conditions which induce vibratory stresses. ~ 12 - 7. The manufacturer of the propeller and the manfaccurer of the engive had recommended inspection periods and time limitations before the accident, 8. ‘the carrier did not follow, nor was {{t required under existing regulations to comply with, manufacturer's recommendations, 9. The FAA was aware of the conditions which {{nduced vibratory stresses, but failed to take timely action to require mandatory compliance with the manufacturers’ recommendations. Z. 10. Airworth’ness directives applicable to the propeller and the engine were issued by the FAA after the accident. ll, No fn-flight vibratory stress tests were conducted on the DH-114 before the supplemental type certificate was {{ssued. 12. Ftetd service coverage and technical fasion between manufacturers and the operators was tnadequate, (b) Probable Cause The National Transportation Safety Board determines that the probable cause of the accident was the separation of the No. I propeller blade of the No. 2 propeller assembly. The blade separated as a result of vibratory stresses which induced fatigue ccacks not readily detectable during routine preflight inspections. Contributing to the acefdent were inadequate overhaul inspection procedures at a certificated repair station and inadequate dissemination and enforcement of recommended mafnteaance practices by the Federal Aviation Administration. 3. TON! As a result of this accident, the Natfonal Transportation Safety Board has issued the following recommendations to the Administrator, Federal Aviation Administration: "Review immediately fts programs for surveillance of eertiffeated repair stattons and its provedures which yovern the granting of supplemental type certificates. ~ J3- "Review its policies relative to users compliance with manufacturers’ service bulletins which may have safety of flight implications, and, where appropriate, issue Airworthiness Directives as soon as possible after service difficulties are discovered." BY THE NATIONAL TRANSPORTATION SAFETY BOARD /s/ WEBSTER B. TODD, JR. Chairman /s/ FRANCIS H. McADAMS Member /s/ ISABEL A. BURGESS Member /s/ WILLIAM R. HALEY Member PHILIP A. HOGUE, Member, did not participate in the adoption of this report. April 14, 1976 ‘ Ps * ~15 - APPENDIX A INVESTIGATION AND DEPOSITIONS 1. Investigation The Miami Field Office of the National Transportation Safety Board was notified of the accident by the Federal Aviation Administration Office at San Juan, Puerto Rico, ac 0722 e.d.t. on July 11, 1975.

[FINDINGS] These failures were typical of damper malfunctions in the engine. Further investigation indicated that damper bushings were not replaced for possibly 4500 hours. Knowing that the blades In the propeller have experienced excessive stresses for many hours, it wovld be impossible to determine the fatigue damage to existing blades and, therefore impossible to know how much more stress the blades can withstand. There is no better and safer method than to replace the blades and start anew. It will eiso re necessary that the Continental Bulletin M68-15 be adhered to - A copy enclosed, This particularly applies to Part 6-18 which {{s encivsed. Further precaution should be taken hy looking at the propellers every 1000 hours. If these procedures are followed, the propellers will have unlimited life safely, After a sertes of tear-downs with good results possibly longer overhaul tiwe can be used. It would also be adviseable'to set the high RPM stop on the governor on a governor test stand and use that setting rather than rely on tuchometers which are inaccurate. We cannot accept the liability of 19 people if the above precautions are rot taken, We will work with you and give you as much financial relief as possible. This whole matter is very serious. Yours truly, HARTZELL PROPELLER, INC. R. V. Grimes President RMG/kas Enclusure

[FINDINGS] Based on visual examination of the fracture suiface in the as-received ecadition with a 10-power glass, the writer beleves the origins of fatigue were at the gouges located at mid-chord and near the trailing edge. Hartzell will submit their report at a later date upon completion of a more detailed examination. R. J. STEINERT Supervising Inspector GL-EMDO-48 -~ 23ALPENDIX E DEPARTMENT OF TRANSPORTATION FEDERAL AVIATION ADMINISTRATION Great Lakes Reeion DATE: MAR 28 1975 300 East Devon Avenue Des Plaines, Illinois 60018 IN REPLY REFER Tu: AGL~210 SUBJECT: Hartzell Model EHC-A3VF-2B/V7636D Propeller Blade Failures; ASJU~265 letters to AGL-214 dated October 10, 1973, Novenber 15, 1974, and February 13 & 27, 1975 Chief, Engineering & Manufacturing Branch, AGL-210 ASO-FSD0-61 (ASJU~265), San Juan, Puerto Rico Attn: Mr. Leonard Davis, Chief This is in response to your referenced letters and is further to our telephone conversation of Feb. 27, 1975, concerning certain service difficulties being experienced with the subject propellers which are used on the Puerto Rico International Airlines’ De Havilland DH-114-2X aircraft (conversions). Enclosed for your information is a copy (Item #1) of our Manufacturing Inspection, AGL-EHM20-48, Service Difficulty Report (Control No. FY75-39), dated Jan. 14, 1975. This report summarizes the results of the examination of Hartzell Model V7636D propeller blade, S/d 65230, submitted by ASJU-255 (Aircraft/ Part Identification and Release PAA Form 8020-2, Nov. 12, 1974). The report findings are self-explanatory. In brief, it is indicaved that the blades failed due to fatigue which originat:d in areas subjected to fcreign object damage. The blade repairs which were accomplished were unsuitable, since the damage which remained resulted in stress raisers thereby contributing to this failure. In reference to your request about the Model V7636D propeller blade, S/N B73741, failure, which was covered in Service Difficulty Report No. 08-31-03-940, dated August 31, 1973 (Ref: yous copy of AGL-210 ltr. to Hartzell dated Oct. 16, 1973), we are enclosing a copy of Hartzell's Engineering Keport No. 411, dated Dec. 17, 1973, for your reference (Item #2). The report, which was submitted to this office in accordance with PAR 21.277, indicates that the blade cracked due to fatigue.

[FINDINGS] Contlrcsccceccvecs APPENDIX H - 30 - May 21, 1975 q Mr. R. V. Grimes President ; I Hartzell Propeller, Inc. E We do not know of any other propeller with a 2,000 hour blade life limitation or such an extensive inspection program. During 743,352 flight hours we have had excellent results. This is supported by the FAA Service Difficulcy Program statistics. Our daily maintenance inspection procedures, excellent experience with the blade and presently having 381 blades on the line and in spares necessitates my request for your personal review of this bulletin. Sincerely, Jumes A, Ceresa Prest{{dent JAC/1 cci- Mr. Rafael ¢€. Ghlestra ‘ Maintenance Manager 3 PRINAIR : APPENDIX I May 30, 1975 Puerto Rican International Atrilines, Inc. International Airport Isla Verde, Puerto Ricu 00913 Attention: James A. Ceresa President Subject: Hartzell Bulletin No. 113, May 13, 1975 Inspection and Replacement of ( ) 7636D Blades and the Inspection of the Blade Clamp Assembly Gentlemen: We received your letter of May 21, 1975, regarding questions concerning the subject Listed above. To understand our position, we must explain the effects of damper bushing wear in the engine to the stresses on the propeller. The Continental 10-520-E engine is equipped with one 4th, one Sth and two 6th orcer dampers. Most of these are necessary to reduce the crankshaft torsional stresses to in turn reduce the propeller vibratory stresses. Damper wear begins early and {{f this wear exceeds possibly a few thousands, the damper {{s no longer effective and the crankshaft torsional stresses increase with propeller stresses {{ncreasing. Both components then experience excessive wear and failures. The propellers sent to us an’ other repair shops indicated obnormal failures and excessive wear. <A letter, enclosed, dated 12/22/70, sent to Prinair suggested precautions you should be taking. Your propellers were not overhauled here after that and no further reports came to our attention. It was assumed that you were adhering to the suggestions in our letter. Page 1 of 2 APPENDIX I - 32 - fuerto Rican International Airlines, Inc. May 30, 1975 Page 2 Finally, further failure reports were received regarding the propellers on your installations. These failures were typical of damper malfunctions in the engine. Further investigation indicated that damper bushings were not replaced for possibly 4500 hours. Knowing that the blades In the propeller have experienced excessive stresses for many hours, it wovld be impossible to determine the fatigue damage to existing blades and, therefore impossible to know how much more stress the blades can withstand. There is no better and safer method than to replace the blades and start anew.

[ANALYSIS] Furthermore, it is apparent that maintenance personnet attempited to correct the high torque of the right engine by repositioning the beta tube which would have resulted in a lower flight idle blade angle than specified by the manufacturer. There was no record showing what exactly was done to readjust the blade angle, and there was no record of a flight idle descent check as required in order to ensure that fuel flows were adjusted properly. It did not appear that the complete procedure was done after the engine change. The Safety Board doubts that a thorough flight idle descent check including accurate fuel flow settings could be accomplished after one 15-minute test flight. It would be highly unlikely to have the fuet flow set correctly after an engine change on the first attempt without having to make a readjustment. The Safety Board believes that had maintenance personnel followed the 33 manufacturer's procedures properly for establishing the flight idle blade angle and flight idle descent fuel flow’ for the airplane, the problems would not have occurred. Consequently, the Safety Board concludes that on the day of the accident, the right engine in the airplane was developing excessive torque at flight idle and that the flight idle hydraulic stop of the propeller could have been set significantly tower than the 7° prescribed by the manufacturer. The significance of an abnormally high flight idle fuel flaw on the right engine becomes evident only when the PLs were retarded to the {{light idle position--a position norrnally used only during descent and the landing flare Under conditions of equal torque on both engines, the propeller blade angles would be essentially equa! in response to propeller governor operation to maintain constant engine RPM. However, when the Pts were retarded to the flight idle position, the right engine would have produced more torque because of the high flight idle fuel flow, and the airplane would have yawed to the left towar«! the engine with the lower torque. At airspeeds above about 70 KIAS, the propeller blade angles would have varied in response to propeller governor operation to maintain constant engine RPM, but the blade angtes on both propellers would have continued to decrease as airspeed decreased. Under these circumstances, had the left engine propeller contacted the flight low pitch stop, in effect, it would have become a fixed-pitch propeller while the right propeller continued to reduce its blade angle in response to decreasing airspeed since the right propeller could have had ar abnormally low flight idle pitch stop setting. Under these conditions, the left propeller would have placed a load on the engine, thereby reducing the engine RPM while the right propeller continued to reduce its blade angle to a lower thrust condition. The left engine goverrior would have sensed an underspeed condition requiring more fuel to increase and maintain the RPM. As more fue! was introduced by the underspeed governor, the let propeller would have produced additional thrust, which in conjunction with the decreasing thrust condition of the right propeller would have caused the airplane to yaw to the right.

[ANALYSIS] Although the right engine and propeller were replaced in early May and the airplane reportedly was test flown satisfactorily on May 5, several discrepancies involving the right engine and propeller continued to be reported thereafter. Later on the same day, the propeller was reported to be operating on beta and NTSing. The toraue was over 40 percent, the airplane yawed, and the flight idle fuel flow was 260 pph. The corrective action taken by the maintenance department to readjust the propeller governor and the FCU underspeed governor would not have corrected the root of the problem. The manufacturer recommends that the two governors be separated by 2.5 percent RPM in order to maintain proper governor control of the engine. Following this corrective action, the airplane was returned to service. On the last flight of the day, the right propeller was reported to have gone into the reverse mode when flight idle was selected on landing. In an interview, the flightcrew was not certain that the propeller went into reverse, but they reported observing the propeller "hunting" (blade pitch oscillation). The maintenance corrective action was to adjust the propeller blade angle. On May 6, eight flights were flawn without a reported discrepancy. However, the first officer who flew the airplane that day confirmed thet the airplane had a tendency to yaw to the left when the PLs were placed in flight idle on landing. On May 7 after the first two flights of the day, the right engine was reported producing 2C percent torque with the PL at flight idle--10 percent more than normal. Again, maintenance personnel readjusted the blade angle on the right propeller. Although the airplane flew eight flights thereafter with no reported discrepancies, the flightcrew that flew the airplane on those subsequent flights reported not having used flight idle until the airplane was on the ground and had slowed down. As a result, it is probable that during these eight flights before the accident, the flightcrew did not encounter a significant yaw problem in the airplane. Considering that the postaccident test of the right engine FCU showed a significantly high flight idle fuel flow of 299 pph, the Safety Board believes that the right engine fuel flow was mis-set well beyond normal limits for operating the C-212-CC. Examination of the FCU showed that at some time, a fuel flow adjustment had been made, and the specific gravity flow of the unit had been changed. Although there was no record to show that the unit had been adjusted by the carrier's maintenance personnel, the Safety Board is inclined to believe that they probably made adjustments in response to the pilot-reported discrepancies. Furthermore, it is apparent that maintenance personnet attempited to correct the high torque of the right engine by repositioning the beta tube which would have resulted in a lower flight idle blade angle than specified by the manufacturer. There was no record showing what exactly was done to readjust the blade angle, and there was no record of a flight idle descent check as required in order to ensure that fuel flows were adjusted properly.

[ANALYSIS] The left engine goverrior would have sensed an underspeed condition requiring more fuel to increase and maintain the RPM. As more fue! was introduced by the underspeed governor, the let propeller would have produced additional thrust, which in conjunction with the decreasing thrust condition of the right propeller would have caused the airplane to yaw to the right. Consequently, the Safety Board believes that the pilot could have been correcting the initial left yaw produced by the right engine with right rudder and encountered a sudcen yaw to the right in conjunction with a rapid reduction in airspeed. A reduction in propeller blade angies to the flight idle low pitch stop would not normally occur in the C-212 until the airspeed is reduced below stall speed and the airplane is on landing rollout. However, a rapid increase in airplane pitch attitude might cause this condition to occur at airspeeds above the airplane's stall speed because the rapid increase in pitch attitude would significantly increase the angle of attack on the propellers momentarily causing the propeller governors to rapidly reduce the blade angles in order to maintain constant engine RPM. The rapid reduction in blade angles coutd, therefore, cause a rapid asymmetric thrust reversal with the left propeller constrained by the low pitch stop and the right propeller blade angle reduced to a lower thrust condition. Moreover, forward movement of the Pls under these circumstances, a conditioned response to a rapid increase in pitch attitude at approach airspeeds, would have aggravated the asymmetric thrust reversal because with the higher blade angle on the left propeller, it wouid have immediately begun producing significantly more thrust than the right propeller with its lower flight idle blade angle. In consideration of all the evidence related to the operation, maintenance, and postaccident condition of the right engine and propeller on the airplane, the Safety Board concludes that the series of trial-and-error maintenance actions performed during the 3 days preceding the accident most likely resulted in the misadjustment of propeller’s flight idle low pitch stop to a degree significantly below the setting prescribed by the manufacturer. Additionally, although the pitot, in an attempt to salvage approach, flew an unstabilized and steep landiny approach that resulted in a rapid increase in pitch attitude at a relatively low altitude that may have aggravated the situation by placing the airplane near an accelerated stall condition, the Safety Board concludes that the pilot's actions probably would have resulted in either a successful flanding or tissed approach had the right engine and propeller been adjusted properly. finally, the Safety Board concludes that the improper maintenance performed on the right engine and propeller during the 3 days preceding the accident relates directly to the cause of the accident. 24 Aircraft Maintenance The Safety Board's investigation disclosed a number of discrepancies in the carrier's maintenance practices. Regarding the maintenance department's actions, the Board believes that there was a failure to troubleshoot the pilot-reported discrepancies correctly. This poor performance led to unnecessary and time-consuming maintenance tasks that aggravated the carrier's maintenance difficulties. The carrier was averaging about seven flights per day per airplane with most flights lasting less than 1 hour based on its flight schedule.

[ANALYSIS] Therefore, the warning time can be significantly reduced or masked depending on the particular maneuver and other environme ital conditions. The Safety Board is concerned that the absence of a stall warning device compromises safety in 14 CFR Parts 135 and 121 passenger-carrying operations by placing too much reliance on a subjectively approved “inherent” buffeting stall warning feature that may be less noticeable during an emergency or in the presence of atmospheric turbuience. There: ure, the Safety Board believes that the FAA should reevaluace tive stall warning certification criteria for airplanes used in Parts 135 and 121 air carrier operations with a view toward requiring stall warning devices on these airplanes. 2.3 Asymmetric Power Condition The Safety Board's investigation determined that other factors also contributed to the pilot losing control of the airplane. Four of the passengers heard unusual engine sounds during the approach, and two ground witnesses remarked about the sound of the engines. The crop duster pilot said the sound was similar to the sound of reverse propeller pitch, and it was associated with the airplane's yaw to the right. These observations indicated that the pilot encountered a problem with an engine or propeller or both. Since no evidence of a failure or malfunction in these components was found, the Safety Board believes the evidence supports the contention that the problem was associated with the manner in which the pilot manipulated the PLs, or with rigging of an engine and propeller controls, or a combination of both. Postaccident metallurgical examination of the pitch change piston from the right propeller estalished that the initiation of the spiral gouge in the wall of the piston corresponded to @ propeller blade angle of 0°. Consequently, the Safety Board concludes that the blades on the right propelier were at an angle of 0° when the propeller dome struck the fence and ground. However, this candition could have been the result of propeller pitch control tinkage distortion when the rigitt enginw separated trom the wing during the impact sequence, of movernent of the PLs ft into the: beta mode by the pilot, or of static misadjustment of the law-pitch stop by maintenance personnel. Although the Safety Board cannot exciude crash dynamics as having produced the low-blade angie on the right propeller, it is of the opinion that distartion of the propeller pitch control linkage did not occur simultaneously with the availability of oil “low from the engine driven pump through the propeller governor to the pitch control cylinder at sufficiently high pressure to force the pitch control piston toward the 0°-blade angle in opposition to feather spring and counterweight forces. Furthermore, unlike the lett engine and propetler which retained its operational integrity for a comparatively long period during the crash sequence, the right engine and propeller was subjected to severe impact forces almost immediately after the right wing tip struck the ground. Consequently, even though components of the right engine and propeller could be functionally tested after the accident, it is probable that the operational integrity of the components was destroyed iminediately after the right propeller struck the ground.

[CONCLUSIONS > FINDINGS] The right propeller flight idle blade angle was inadvertently mis-set to operate below prescribed timits. The weather wa3 not a factor in the accident, The airport facilities, personnel, and equipment involved operated normally and were not factors in the accident. The pilot made an unstabilized approach at a relatively steep angle and at a high rate of descent resulting in the airplane sinking below a normal approach path to the runway. The pilot experienced an asymmetric power condition as the airplane was slowed rapidly in his attempt to arrest a high rate of descent and to salvage the approach or to initiate a go-around maneuver. The retracted position of the flaps could have been a factor in the accident and could have contributed to a stall. The pilat lost control of the airplane at an altitude from which recovery could not have been accomplished. Standards for inherent aerodynamic qualities are probably not adequate for providing stall warning in new aircraft used in new aircraft used in 14 CFR Parts 135 and 12i-passenger-carrying operations. The impact forces and the destruction of the cockpit made the accident unsurvivable for the flightcraw. The impact forces were survivable for the passengers because the structural integrity of the cabin was maintained, the seats remained attached to the floor tracks, and passengers used their seatbelts which remained intact. The carrier's management and supervision of the maintenance department personnel was deficient. The FAA FSD0's Initial certification of the carrier was deficient in the area of maintenance recordkeeping. The bilateral type certification program of the CASA C-212 was not managed effectively by the FAA. The reorganizational changes, personnel changes, and the limited availability of resources within the engineering and operations departments of FAA are cortributing factors. 3.2 Probable Cause The National Transportation Safety Board determines that the probavle cause of the accident was improper maintenance in setting propeller flight idle blade angle and engine fuel flow resulting in @ loss of contral from an asymmetric power condition. A factor contributing to the accident was the pilot's unstabilized visual approach. 4, RECOMMENDATIONS As a result of the Fischer Bros. accident in Romulus, Michigan, on March 4, 1987, the Safety Board issued Safety Recommendations A-87-27 and -28 to the FAA: A-87-27 Issue a gerreral notice (GENOT) immediately to all U.S. owners and operators of the CASA €-212 airplane describing the background and significant findings of the recent flight test of the CASA C-212. The notice should provide an evaluation of the exitting CASA C-212 stall characteristics, operational precautions, and training procedures to prectude inadvertent stalls until an approved artificial stall warning system is installed. A-87-28 Expedite the rulemaking action to require installation of an artificial stall warning system on the CASA C-212 airglanes. On June 16, 1987, the FAA responded to these recommendations stating that for Safety Recommendation A-87-27, the FAA had issued a GENOT to al! flight standards field offices addressing the flight characteristics of the CASA 212 aircraft.

[CONCLUSIONS] The captain did not take the necessary steps to ensure that the first officer achieved the 300-knot or greater airspeed required for the windmill engine restart procedure and then did not demonstrate command authority by taking control of the airplane and accelerating it to at least 300 knots. Conclusions 71 Aircraft Accident Report 11. The first officer’s limited experience in the airplane might have contributed to the failed windmill restart attempt because he might have been reluctant to command the degree of nose-down attitude that was required to increase the airplane’s airspeed to 300 knots. 12. Despite their four auxiliary power unit-assisted engine restart attempts, the pilots were unable to restart the engines because their cores had locked. Without core rotation, recovery from the double engine failure was not possible. 13. The General Electric CF34-1 and CF34-3 engines had a history of failing to rotate during in-flight restart attempts on airplanes undergoing production acceptance testing at Bombardier. 14. Both engines experienced core lock because of the flameout from high power and high altitude, which resulted from the pilot-induced extreme conditions to which the engines were exposed, and the pilots’ failure to achieve and maintain the target airspeed of 240 knots, which caused the engine cores to stop rotating; both of these factors were causal to this accident. 15. The importance of maintaining a minimum airspeed to keep the engine cores rotating was not communicated to the pilots in airplane flight manuals. 16. The captain’s previous difficulties in checklist management, the situational stress, and the lack of simulator training involving a double engine failure contributed to the flight crew’s errors in performing the double engine failure checklist. 17. The pilots’ failure to prepare for an emergency landing in a timely manner, including communicating with air traffic controllers immediately after the emergency about the loss of both engines and the availability of landing sites, was a result of their intentional noncompliance with standard operating procedures, and this failure was causal to the accident. 18. The pilots’ unprofessional operation of the flight was intentional and causal to this accident because the pilots’ actions led directly to the upset and their improper reaction to the resulting in-flight emergency exacerbated the situation to the point that they were unable to recover the airplane. 19. Revised high altitude training syllabuses for pilots who operate regional jet airplanes would help ensure that these pilots possess a thorough understanding of the airplanes’ performance capabilities, limitations, and high altitude aerodynamics. 20. Because most training for stalls occurs with the airplane at low altitudes, the training methods may introduce a bias in stall recovery techniques by encouraging pilots to minimize altitude loss and not fully recognizing other available recovery techniques. Conclusions 72 Aircraft Accident Report 21. Additional training might improve pilot response to stickpusher activation, but such training, if not provided correctly, could have an adverse impact on existing stall recognition and recovery protocols. 22.

[ANALYSIS] CVR evidence showed that the flight crew recognized the nature of the emergency; specifically, about 2155:23, one pilot stated to the other, “we don’t have any engines.” However, the CVR also showed that the flight crew did not begin the first item of the double engine failure checklist (which is a memory item) until 1 minute 19 seconds after the statement on the CVR recording about the double engine failure. Also, FDR data showed that the pilots did not achieve and maintain the target airspeed of 240 knots (another memory item on the checklist). Because the flight crew did not achieve this airspeed, both engines’ cores had decelerated and stopped before the airplane descended through an altitude of 28,000 feet. The double engine failure checklist indicated that, between the altitudes of 21,000 and 13,000 feet, the windmill restart procedure should be used to relight the engines. This procedure required that the pitch attitude of the airplane be reduced to and maintained at -8º to accelerate the airplane to an airspeed of 300 knots or greater. However, the pitch inputs made by the flight crew were not of sufficient magnitude and were not sustained. As a result, the crew did not achieve the 300-knot or greater airspeed required for the procedure; FDR data showed that the highest airspeed attained during the restart attempt was 236 knots and that the engines’ N2 (core rotation) indications remained at zero. The Safety Board concludes that the captain did not take the necessary steps to ensure that the first officer achieved the 300-knot or greater airspeed required for the windmill engine restart procedure and then did not demonstrate command authority by taking control of the airplane and accelerating it to at least 300 knots. The Safety Board further concludes that the first officer’s limited experience in the airplane might have contributed to the failed windmill restart attempt because he might have been reluctant to command the degree of nose-down attitude that was required to increase the airplane’s airspeed to 300 knots. Because the flight crewmembers were unsuccessful in their attempt to relight the engines using the windmill procedure, they elected to descend the airplane to an altitude of 13,000 feet so that they could attempt to restart the engines with the APU. Once the airplane descended to an altitude of 13,000 feet, the flight crew attempted four APU-assisted engine restarts (two attempts per engine) during a 5-minute period (2207:04 to 2212:07) and between the altitudes of about 12,900 and about 5,000 feet. FDR data showed that the N2 indications for both engines remained at zero during the restart attempts. The Safety Board concludes that, despite their four APU-assisted engine restart attempts, the pilots were unable to restart the engines because their cores had locked. Without core rotation, recovery from the double engine failure was not possible.105 105 Section 2.2.3.1 discusses the flight crew’s performance of the double engine failure checklist. Analysis 49 Aircraft Accident Report The Safety Board considered whether engine hardware was a factor contributing to the lack of core rotation.

[ANALYSIS] Specifically, the accident airplane’s engines flamed out from high power and high altitude, whereas the engines installed on the production airplanes were shut down only after their internal temperatures were stabilized.108 Flameouts at high power and high altitude produce even greater thermal distress because internal temperatures are the hottest at high power settings and the air is colder at high altitudes. The increased thermal shock exacerbates the loss of component clearance 106 The FDR parameter for the LCV position shows that the valve is open once it reaches a position that is greater than 4.8º from its normally closed position. 107 The FDR recorded a steady decline in engine oil temperature after the upset and showed that this steady decline ceased at the time of the engine restart attempts. 108 On November 20, 2006, the Safety Board issued recommendations to the FAA to address this and other core lock-related issues; see sections 1.18.3.1 and 4.3 for information. Analysis 50 Aircraft Accident Report and alignment. Because the accident engines flamed out under these conditions, axial misalignment caused the seal teeth, which were positioned aft of their normal operating grooves, to contact stationary abradable material when radial seal clearances closed down. Once core rotation stopped, binding prevented core rotation from resuming during the windmill and APU-assisted restart attempts. No physical evidence of core lock was found inside the engines because the thermally induced interference occurred after core rotation had stopped and operating clearances were restored afterward as the engine cooled. The Safety Board concludes that both engines experienced core lock because of the flameout from high power and high altitude, which resulted from the pilot-induced extreme conditions to which the engines were exposed, and the pilots’ failure to achieve and maintain the target airspeed of 240 knots, which caused the engine cores to stop rotating; both of these factors were causal to this accident. 2.2.3.1 Performance of the Double Engine Failure Checklist A critical error made by the flight crew while performing the double engine failure checklist was failing to establish and maintain an airspeed of 240 knots before beginning the procedures to restart the engines. During the public hearing on this accident, a GE manager testified, “as long as core rotation is maintained, you will not have core lock … we have a body of data that shows that 240 knots maintains core rotation.” Another critical error made by the flight crew was failing to increase the airspeed to at least 300 knots before beginning the windmill restart procedure. About 2159:16, when the FDR resumed operation after a loss of power lasting about 4 minutes, the airplane’s airspeed was 178 knots. About 2200:38, the captain told the first officer to increase the airspeed to above 300 knots, and the first officer acknowledged this instruction. About 1 minute later, the captain again told the first officer to increase the airspeed to 300 knots. However, FDR data showed that the maximum airspeed achieved during the procedure—236 knots—was achieved only briefly.

[ANALYSIS] Analysis 53 Aircraft Accident Report The company’s double engine failure procedure stated that, if neither engine was restarted, the flight crew should consider a forced (emergency) landing. Five airports that were suitable for an emergency landing were located within 60 miles of the upset location (see section 1.16.1.3). However, the flight crew did not consider making an emergency landing at any of these airports. Between the time that the flight crew first realized that a double engine failure had occurred and the time that the captain notified the controller of the actual situation, the airplane had descended from 35,000 to 10,000 feet. When the airplane was at an altitude of 10,000 feet, only one airport (AIZ) was still within the airplane’s best glide range, but the pilots had already overflown that airport. At that altitude, JEF was just outside the airplane’s best glide range.113 The aircraft performance study for this accident showed that drag produced by the 5.6º spoiler deployment had a minimal effect on the airplane’s ability to reach JEF. A double engine failure in a two-engine airplane is an emergency situation that requires timely action to stabilize the situation and effective coordination of all available resources and options to ensure a successful outcome. The flight crew’s failure to discuss an emergency landing during most of the descent demonstrated the crew’s poor judgment. Also, the captain’s failure to inform the controllers about the true nature of the situation early in the descent demonstrated the pilots’ apparent reluctance to have their deviation from standard operating procedures and their unprofessional behavior detected. Multiple opportunities existed during the descent for the pilots to talk with controllers and between themselves about the need to make an emergency landing. The pilots acted in a manner that was not consistent with ensuring safety of flight or effectively managing an in-flight emergency. The Safety Board concludes that the pilots’ failure to prepare for an emergency landing in a timely manner, including communicating with air traffic controllers immediately after the emergency about the loss of both engines and the availability of landing sites, was a result of their intentional noncompliance with standard operating procedures, and this failure was causal to the accident. 2.2.5 Accident Sequence Summary This accident was not the result of a single action, decision, or mistake by the flight crew. Instead, the accident was the result of poorly performing pilots who intentionally deviated from standard operating procedures and basic airmanship. Also, CVR evidence showed that the pilots’ tone in the cockpit was unprofessional114 and that their attitude was inconsistent with the demands associated with, and the precision required for, flying a high performance turbojet airplane. 113 On the basis of the airplane’s location at this point in the flight, the air traffic controller correctly directed the airplane to JEF, which was essentially straight ahead along the airplane’s flightpath. If the airplane had been directed to AIZ, a 160º left turn would have been required, which would have cost the airplane altitude, airspeed, and distance. 114 As indicated in the CVR transcript in appendix B, the flight crew’s conversation was generally casual, and sounds of joking and laughing were heard during the recording.

[ANALYSIS] The Safety Board concludes that, despite their four APU-assisted engine restart attempts, the pilots were unable to restart the engines because their cores had locked. Without core rotation, recovery from the double engine failure was not possible.105 105 Section 2.2.3.1 discusses the flight crew’s performance of the double engine failure checklist. Analysis 49 Aircraft Accident Report The Safety Board considered whether engine hardware was a factor contributing to the lack of core rotation. However, engine teardowns found no mechanical failures or evidence of any condition that would have prevented engine core rotation. The Safety Board then considered whether an electrical or mechanical problem interfered with the ability of the cores to achieve rotation. However, inspection and testing of the APU and other airplane start system components found no conditions that would have prevented torque delivery to either engine during the four APU-assisted restart attempts. Also, FDR data showed that the LCV was open during these restart attempts,106 and testing showed that it was capable of providing pneumatic power from the APU to start initial rotation in each engine. A systems simulation study (see section 1.16.4) showed that the LCV had likely opened by at least 18º during the flight, which should have resulted in some detectable core rotation if the engine cores were capable of rotation. In addition, the change in engine oil temperatures107 showed that air was flowing through the ATS. This finding, along with the results of the airplane start system component testing, showed that the LCV was likely fully open. The Safety Board considered whether the slight increases in core-driven hydraulic pump pressures shown on the FDR during the final minutes of the flight (as described in section 1.16.3) were evidence of engine core rotation. The airplane’s start system was operating normally during the restart attempts; thus, if the cores were free to rotate, they would have quickly accelerated. As a result, the Board determined that the slight increases in hydraulic pump pressures that were recorded on the FDR were not the result of engine core rotation. On the basis of the information discussed in section 1.18.2, the Safety Board concludes that the GE CF34-1 and CF34-3 engines had a history of failing to rotate during in-flight restart attempts on airplanes undergoing production acceptance testing at Bombardier. GE attributed the problem to interference contact at an air seal in the high pressure turbine and, along with Bombardier, referred to this condition as core lock. The lack of core rotation on the accident airplane engines was similar to the instances of core lock experienced by CF34 engines during Bombardier’s acceptance testing, except that the accident airplane engines were exposed to more severe thermal distress than the engines on the production airplanes. Specifically, the accident airplane’s engines flamed out from high power and high altitude, whereas the engines installed on the production airplanes were shut down only after their internal temperatures were stabilized.108 Flameouts at high power and high altitude produce even greater thermal distress because internal temperatures are the hottest at high power settings and the air is colder at high altitudes.

[ANALYSIS] Wichdrawal probably was the inadvertent result of the manner in which the power and high-pressure fuel valve levers were manipulated during the crew's attempt to feather the left propeller. Again, however, a datermination of the precise sequence of events which caveed the withdrawal is ivi possible from the avaliable evidence. Performance data indicates that with the laft propeller wind« milling at « 16° blade angle, Level flight was not possible at 125 note, aven with gear and fiaps retracted and full power, including waier/methanol tnteccion end 100 percent fuel trim, on the other engine, In addition, with the asymmetric thrust condition further aggravated by tha windmilling left propeller, control of the aircraft would heve been difficult, In thie instance, it appears that the right engine «= up until seconds before impact, when the powar lever was retarded -- was operating at 15,000 repem, 16/ and the gear and flaps were retracte!, ‘"1e fuel trim setting remained at zero, and the airspeed was approximately 120 knots. In view of tha foregoing and the low altitude at which the Bpooldown occurred (700 feat or 450 feet above ground level), it appears that the accident became inevitable when the left engine was shu’: down but the propeller failed to feather, A considerable investigative effort was expeniad to determine the possible reasons undarlying the failure of the left propaller to feather. Ae noted above, all operating components of tha feathering system, daciuding the propeller contro}} unit with the worn cam follower roller, were thoroughly tested, and all performed cheir prescribed functions, It is possible that a mistigging of the propeller control unit system or high-pressure fuel valve. lever, riot detectable from the posterash examination or testing of thene componente because of thety condition, could have accounted for the unsuccessful attempts to feather the left engine. It is further possible that a single discrepancy, perhaps of a transient nature, was responsible both for tha failure to feather and for the cruise pitch lock hangup, inasmuch ap there are som ponents which are common both to tha feathering and tha pitch lock with» drawal systems, However, any such discrepancy ramaine only e@ possibility, unsupported by any positive evidence, I a FOLEY Cees Rae a ay oe It is also possible that either the captein or the first of fi» cer failed to place the high-pressure fuel valve lever to the feather position but rather placed that lever in the closed position, which could account for the fact that the engine shut down but did not feather. ie] the position was determined by means of an anergy gradient analyale and the fact that the right spill-valve was found in the open poxsltion, LOE a ELIE NITES EON FOE TOME ET. ele eT gc tlc iataet + * on 27 wi In thie connection, we note that, to enter the feather position, the high-pressure fuel valve lever must be moved past a detent.

[ANALYSIS] The calculated takeoff weight for N7818M was 45,233 pounds. Tha center of gravity (c.g.) wav at 25.6 percent of the maan aerodynamic chord (MAC). Takeoff ¢.g. limits were from 20 to 35 percent MAC, The estimated crash weight of the aircraft was 43,343 pounds, with ac.g. at 25.6 percent MAC, The c.g. Limite were from 19 to 25 percent MAC at that weight. 2. Engine ard Propeller Data N7818M was powered by two Rolls-Royce Dart 53267 jet turbine engines, aach equipped with a Dowty Rotol R=257/4=30-4/60 propeller. The engines were each rated at 1,990 ehaft horsepower with water/methanol injection, and 1,910 maximum to 1,835 mininum shaft horsepower without injection. Engine and propeller specifics are as follows: Engines Position Serial No. Time Since Overhaul (780), No. 1 14057 1,989326 No, 2 13994 3339320 Propellers Position Serial No. TSO Total Time I 2 No. DRG 163/66 2 533303 6,097 340 No. DRG 451/66 3,494 :31 9.224357 Propeller Pitch Locks Serial No, TSO 163/66 2 259258 409/66 3, 988330 ~ Fe a go x ” Tyg RE ~ piers ON te UN RN tt UY Peet OR cee are woes ee “ ; j ; j j i t $ i {{ i e 40 « APPENDIX C, Page 2 The left propeller control unit, serial No. 1314/66 had accumu lated 2,491 hours since che last overhaul on September 17, 1970, on Detober 24, 1970, after 210217 hours of operation, the unit was returned to the overhaul facility for repair. ‘The fuel valve lever was reported to be binding when the unit was hot. The repair was completed and the unit was returned to Service on the left engine on July 2, 1972, No writeups wera recordad regarding thie unit after that data, Eun22 78 Powerplant Information Ip tha FH-22 73 ingtallation, each engine is equipped with a constant speed, hydraulically Operated, feathereble, four«bladed propeller, Control of tha powerplant is accomplished manually by a system of cables and push-pull roda between the cockpit angine control quadrant and the eagine nacelles.

[ANALYSIS] Before the flights in N7818M on March 3, 1972, there were no recorded discrepancies on either powerplant which would indicate that they were incapable of normal operation. If it is assumed that the automatic feathering circuits were checked by the flightcrew invelved in this accident, in accordance with Mohawk procedures, then 211 associated components functioned satisfactorily at that time, since there ia no indication to the contrary. Examination of the engines disclosed that the left engine was not operating at impact and that the left propeller was rotating at a very low r.p.m. The right engine was operating at a Low power setting and rotating at less than 11,000 r.p.m., which ie most Likely attributable to a power retardation several seconds before impart, the right propeller-blade angle was determined to have teen about 16° e: impact, which conforms to the expected angle with an engine r.p.m. of less than 11,000 and with the flight fine pitch lock extended, Tha left propeller-blade angle also was datermined to have been at 16° at impact, thus indicating that at some point before the crash, the blade angle passed through the 28° position where it apparently was hung up for gome pertod during the approach. Extensive laboracory tests were mada of powerplant components related to propeller operation, Both propeller control units were axamnined and functionally tested, and both were found capable of normal operation, The worn cam follower roller in the left propeller control unit did not affect normal functioning of the propeller control unit. The only malfunction that could ba induced occurred when the propeller control unit feathering lever was moved slowly from the run-auto position into the lockout position while the cruise pitch lock withdrawal solenoids were inoperative, which produced intermittent pressure fluctuations in the coarse, fine, and inlet oil Lines, accompanied by 4 flickering cruise pitch Lock light. mit Laie a ver : Sea enh none Sw ome hy pa tN. » 22 » The left propeller control unit was extensively flight tested and it performed repeatedly the selected functions, including the emergancy~out and feather functions, without fail, delay, or malfunction, the left propeller control unie rigging was extensively damaged. However, 4i1 control rods and control rod ends were accounted for, and there was no evidence of a gross misrigging of these controls. The feathering units from both propellers were examined and functionally tested. Both operated normally. The left feathering unit was flight.tested, with wo indications of abnormal performance.

[ANALYSIS] Both high-pressure fuel valva levers were in the emergencyeout posic.on, and both power levers were in the off poet« tion, 14/ The CVR affords little additional evidence about what was happening in the cockpit at this time. After the captain announced his decision to ehut down the lefé engine, the series of popping sounds continued (until 2047:18)3; thus indfieating that his efforts to withdraw the pitch lock continued. At 2047220, or 2 saconds after the last popping sound, the sound of engine spooidown began, It is appare ent, however, that although the fuel to the left engine was shut off, thea attempts to feather the propeller were unsuccessful, and it was left in a windmiliing condition, The deceleration curve recordad by the CVR was riot similar ¢o the test decaleration curves for feathering deceleration, but rather indicated that the engine rotating speed da« erensed to a value of approximately 6,500 r.p.m, Moreover, the left propeiler was determined to have been rotating at a vary low 7.p.m, at impact. The problems resulting from a windmilling propeller, and tha associated high degree of drag, were inmediately manifaated to the crew. The rate of descent of the aircraft tncreased, and (at 2047 336) the captain asked for assistance on the rudder pedal. At 2047338, the captain made a commant that reflected his diffleulty in attempting to feather the propeller, which commant could ba further interpreted as a request for the first officer's assietance. At 2047:51, the captain inatructed the first officer to "tell ‘em we're gonna land short, we're in trouble," which was relayed to the concroller seconds later. At 2048307, the aircraft crashed. 15/ IA] Any significance attached to the position of these various controls must be viewed in light of tha possibility that they could have been movad by impact forces or during the extraction of bodies from the cockpit. SSAC cei AR oY MRT te Neat a orm mo a NY om During the period between angine shutdown and impact, there were 4 number of clicks, In pairs, racorded on tha CVR, Howaver, none of these could be identified with a particular control or ewitch, and any anatyela thereof would primarily involva speculation, 6 de ged oy te et “s OR mee IE 4: At some time during the attempte to feather the left propeller, the cruise pitch lock withdrew, which allowed the blade angle to de« crease to the 16° position, where it was constrained by the flight fine pitch lock. Wichdrawal probably was the inadvertent result of the manner in which the power and high-pressure fuel valve levers were manipulated during the crew's attempt to feather the left propeller.

[ANALYSIS] An analysis of performance data showa that, with both engines running at about 12,000 r.p.m., 17/ and with the Lest propeller blades constrained at an angle of 28°, thrust was aval lable, on demand, that would have sustained a climb gradient, even with the drag brake extended and zero fuel trim, It 18 therefore apparent that, ij77 thie figure was derived from the spactvographic analysis of engine sounds recorded by the CVR, RP et eerie lS a ett ea ee oy - » 23 a prior to engine shutdown, the descent of the airoraft conld have been arrested, level flight smintained, or a climb commenced at any time the captain chose to use the excess thrust avatlabla, Tt ig also obvious that if the gear or dvag brake were retracted, and/or the fuel trim were set at 100 percent, tha performance capa bilities would have been even further enhanced, ‘the fuel trim apparent« ly was not moved above the zero position beceuse such a change te normal ly not accomplished until after the landing gear is extended during the Before Landing Chacklist, and the orew on this flight never reached a point where that portion of the checkitet was called for. 18/ The drag brake was clearly down during that portion of the approach pre» ceding the leveloff at 1,750 feet, based on the combination of decreas ing airspeed and steady rate of descent. ‘Tha position of the drag brake thereafter is less clear, although, based on the energy gradient analysis, it appears likely that it was extended during most or all of the period until the left engine was shut dom. In any event, the Board coneludes that the manner in which fuel trim and the drag brake were handled, al» though not the best practice under the circumstances, was not a causal factor because: (a) bafore engine shutdown, adequate thrust was avail» able to climb, even with fuel trim set at zero and drag brake down, and {{b) after engine shutdown and the concomitant failure of the left proe pailer to feather, placement of tha fual trim to 100 percant ant rae traction of the deag brake, the lattar ¢* which was actually accomplished, would not have avoided the crash, After discounting the abov: «diacussed possibilities, and weigh ing all of the pertinent evidence, tha Board is convinced that the con« tinued descent below the approach mintmum altitudes was directly related to the ecew's preoccupation with the cruise pitch lock malfunction and the associated lack of coordination and eifective task sharing in deal« ing with that problem.

[FINDINGS] The flight crew’s continued descent to Philadelphia International Airport was not 4. inappropriate given that there was no evidence of abnormalities other than the odor and that no cockpit alerts had activated. The increased airflow that resulted from the Fumes Evacuation checklist actions 5. diluted the smoke and inhibited its detection by either the smoke detection system or flight crewmembers and provided the fire with additional oxygen. The aviation industry initiative on smoke, fire, and fumes provides specific guidance 6. on when and how flight crews should respond to evidence of a fire in the absence of a cockpit smoke and/or fire warning. The fire on board the accident airplane initiated as a smoldering fire. 7. The fire was detected by the airplane’s smoke and fire detection system after the fire 8. breached a cargo container, at which time, it proceeded to spread, and the growth of the fire after landing was fed by air entering through open doors and burnthrough holes. The exact origin and cause of the in-flight fire on board the airplane could not be 9. determined due to the destruction of potentially helpful evidence; however, the available evidence suggests that the fire most likely originated in container 12, 13, or 14. The current certification test standards and guidance for smoke or fire detection 10. systems on board many aircraft are not adequate because they do not account for the effects of cargo and cargo containers on airflow around the detection sensors and on the containment of smoke from a fire inside a container. Conclusions National Transportation Safety Board A I R C R A F T Accident Report 65 The threat from cargo fires could be mitigated by the installation of fire suppression 11. systems. Flight crews on cargo-only aircraft remain at risk from in-flight fires involving both 12. primary and secondary lithium batteries. The emergency response for this accident was timely. 13. Some aircraft rescue and firefighting personnel are not adequately trained on the 14. use of the high-reach extendable turret with skin-penetrating nozzle, reducing the effectiveness of the device in fighting interior aircraft fires. Philadelphia International Airport aircraft rescue and firefighting personnel were not 15. familiar with the accident airplane’s main cargo door, which adversely affected their ability to access the airplane’s interior to fight the fire. The availability of accurate and complete airplane diagrams would improve aircraft 16. rescue and firefighting personnel’s knowledge and familiarity with fleet configurations and would facilitate emergency response operations. A floor level emergency exit, including one equipped (when appropriate) with an 17. evacuation slide, would enable more efficient emergency egress for airplane occupants than cockpit window exits, and the associated, instructional placarding of such an exit would assist emergency responders with locating and operating the exit door and accessing the interior of the airplane. United Parcel Service Company (UPS) guidance on hazardous materials information 18. retrieval and dissemination was inadequate, which resulted in UPS personnel not providing emergency responders with detailed information about the hazardous materials on board the airplane in a timely manner. The requirements of 49 19.

[ANALYSIS] However, the flight engineer saw no smoke either of the two times that he visually checked the main cargo compartment. This evidence indicates that the fire initially did not generate a significant amount of smoke and was most likely initiated as a smoldering fire inside a cargo container. The construction of the cargo containers, which results in restricted airflow in or out of the container, likely inhibited the growth and detection of the fire in its initial stages. On the basis of this evidence, the Safety Board concludes that the fire on board the accident airplane initiated as a smoldering fire. Once the fire breached the cargo container in which it initiated, it would have begun to spread to adjacent containers. Detection of the fire by the main cargo compartment smoke detectors most likely occurred around the time of the first container burnthrough. The smoke detector in lower cargo compartment 33 alerted about 1 minute after the main cargo compartment alert, and some of the captain’s displays then began to falter, indicating the continued progression of the fire. The flight engineer first saw smoke when he exited the cockpit to close the main cargo air shutoff valve and black smoke emanated from the valve’s access panel. About 2 minutes later, almost immediately after touchdown, the flight engineer reported that smoke had begun entering the cockpit. The smoke continued to worsen after the airplane came to a stop, and the smoke in the cockpit became so thick that the two pilots could not see each other before evacuating the airplane. ARFF personnel who entered the airplane through the L1 door observed smoke but no fire in the main cargo compartment. Flames were first observed about 40 minutes after the airplane landed when ARFF personnel opened the right-forward overwing hatch Analysis National Transportation Safety Board A I R C R A F T Accident Report 50 and noticed flames above the containers just aft of the opening. The continued growth of the fire after the airplane landed was probably affected by the introduction of fresh air through the openings in the airplane, both through its open doors and eventually through holes created in the fuselage as the fire ultimately burned through the fuselage crown. The growth of the fire was also affected by the combustible nature of the packages and packing materials in the cargo containers, which provided a readily ignitable fuel source for the growing fire. The fuselage burned through aft of the wings (near containers 12, 13, and 14) about 2 hours after landing. On the basis of this evidence, the Safety Board concludes that the fire was detected by the airplane’s smoke and fire detection system after the fire breached a cargo container, at which time, it proceeded to spread and that the growth of the fire after landing was fed by air entering through open doors and burnthrough holes. Origin and Cause of Fire 2.4  Despite the length of time that the fire burned and the resulting destruction of potentially helpful evidence, the postfire condition of the cargo containers and contents and the surrounding airplane structure were examined for any evidence indicating the location from which the fire initiated.

[ANALYSIS] Origin and Cause of Fire 2.4  Despite the length of time that the fire burned and the resulting destruction of potentially helpful evidence, the postfire condition of the cargo containers and contents and the surrounding airplane structure were examined for any evidence indicating the location from which the fire initiated. As a result of these examinations, the lower cargo compartments were eliminated as the origin of the fire because they sustained no heat‑related damage, and the outer surfaces of their ceiling liners only exhibited sooting and smoke damage. The fire/smoke detector activations in lower cargo compartments 33 and 34 resulted from smoke infiltrating the lower compartments from the main cargo compartment above this area. Cargo containers 1 to 11 were eliminated as the fire initiation location because they sustained minimal to moderate thermal damage and the surrounding fuselage sustained little to no structural damage. Further, the ceiling liners were still in place for containers 1 to 7. Although the ceiling liners in containers 8 to 11 sustained thermal damage, it most likely resulted after the overwing hatches were opened. Containers 15 and 18 were eliminated as the fire initiation location because the cargo in these containers was accounted for during the on-scene examination and no evidence of ignition sources was found. The contents in containers 16 and 17 were heavily damaged, and fuselage burnthoughs were located near both containers. However, containers 16 and 17 were eliminated as the fire initiation location, in part, because the aftward movement of air in flight and on the ground would have made it unlikely that the fire started in either of these containers. Further, the amount of damage sustained near cargo container 17 might have resulted when ARFF personnel opened the L4 and R4 doors soon after they arrived on scene. ARFF personnel indicated that no fire was observed inside the rear of the airplane at this time. Pictures taken soon after the response began do not show any visible fire in that area, but pictures taken 2 hours into the emergency response show this area heavily involved in fire. Analysis National Transportation Safety Board A I R C R A F T Accident Report 51 Conversely, ARFF personnel did report first observing flames inside the airplane in an area near containers 12 and 13, and this area is also the area above where the first smoke detector activated in the lower cargo compartments. Further, the amount of damage to containers 12, 13, and 14, their contents, and the areas surrounding these containers, including two complete and a partial fuselage burnthrough, indicates that the fire most likely originated in one of these cargo containers. The lowest point of fuselage damage, which could be indicative of the point of origin of the fire, was located in the right aft portion of container 12 near container  13, where the damage extended down to the cargo compartment floor. Container 14 exhibited heavy damage to the contents and was relatively near the areas of the fuselage with the most damage.

[ANALYSIS] The lowest point of fuselage damage, which could be indicative of the point of origin of the fire, was located in the right aft portion of container 12 near container  13, where the damage extended down to the cargo compartment floor. Container 14 exhibited heavy damage to the contents and was relatively near the areas of the fuselage with the most damage. In addition, a large number of the items in this container were not identified either during the on-scene examination or through the shipper interview calls. As a result, container 14 could not be eliminated as a source of the fire. No evidence of explosion or high-temperature fire (for example, melted steel components) was found. Several factors eliminated the airplane’s wiring as a possible source of the fire, including the lack of any electrical burning odor, the lack of fused wires or damage associated with arcing wires, and the lack of system or display anomalies until almost 20 minutes after the flight crew first detected an odor. Other than secondary lithium batteries, which are discussed below, all of the recovered declared and undeclared hazardous materials were eliminated as ignition sources because they were found intact and relatively undamaged. No additional hazardous or questionable items (that is, undeclared hazardous, chemical, or flammable materials) were identified during the identification effort. Examinations of the cargo containers and their contents revealed that a number of electronic devices, such as laptop computers, contained some type of power source, specifically secondary lithium cells. All electronic devices and battery packs and cells that had any type of thermal damage were examined on scene; however, most of the items were too damaged for further documentation. Several laptop computers, loose battery cells, and battery packs with severe damage were retained and sent to the Safety Board’s laboratory for further examination. During the cargo examinations, no batteries were found that exhibited any damage identifying a source of ignition. Because the fire most likely originated in cargo container 12, 13, or 14, investigators attempted to determine the contents of the packages that were not accounted for on scene by contacting the shippers of the packages in these containers. This effort was unable to determine the contents of all of the packages in these containers; however, the effort did reveal that several electronic devices likely containing secondary lithium batteries were shipped in these containers. Unfortunately, the lack of information about the devices or the batteries prevented any determination of whether these batteries were associated with previously known recalls. The Safety Board concludes that the exact origin and cause of the in-flight fire on board the airplane could not be determined due to the destruction of potentially helpful evidence; however, the available evidence suggests that the fire most likely originated in container 12, 13, or 14.

[ANALYSIS] Analysis National Transportation Safety Board A I R C R A F T Accident Report 46 Analysis 2. General 2.1  The flight crewmembers were properly certificated and qualified under Federal regulations. No evidence indicated any preexisting medical or physical condition that might have adversely affected the flight crew’s performance during the accident flight. Although overnight flight schedules can be tiring and produce cumulative fatigue over successive nights,67 the accident occurred on only the second night of the trip sequence, and all of the flight crewmembers had made use of sleep opportunities during the day before the flight. Further, the accident occurred less than 2 hours into the trip, about 0000, a time at which neither pilot typically slept.68 Finally, there was no evidence of performance deficiencies that were clearly discernible and consistent with the known effects of fatigue. Therefore, the Safety Board concludes that no evidence was found indicating that fatigue degraded the performance of any of the flight crewmembers on the day of the accident. Examinations of the recovered components revealed no evidence of any preexisting powerplant, structural, or system failures. This analysis discusses the accident sequence, including the flight crew’s performance; the quality of fire and smoke emergency checklists; smoke and fire detection and suppression systems on cargo aircraft; cargo aircraft smoke and fire detection systems test certification requirements; ARFF training inadequacies; cargo airplane emergency exit requirements; hazardous materials information dissemination procedures; and issues related to the transport of lithium batteries on cargo airplanes. Flight Crew Performance 2.2  Actions During Descent and Landing Sequence 2.2.1  The accident flight was uneventful until just after the descent to PHL began, about 2335, at which time, the first officer detected an odor and asked the captain and flight engineer if they smelled anything. The captain stated during postaccident interviews that 67  P.H. Grander, K.B. Gregory, L.J. Connell, R.C. Graeber, D.L. Miller, and M.R. Rosekind, “Flight Crew Fatigue IV: Overnight Cargo Operations,” Aviation, Space, and Environmental Medicine, Vol. 69 (1998), pp. B26‑B36. 68  When off duty, the captain typically went to bed between 0100 and 0200 and the first officer between 0000 and 0100. Analysis National Transportation Safety Board A I R C R A F T Accident Report 47 he evaluated diversion alternatives at this time but decided to continue to PHL. The flight crew’s decision to continue to PHL is discussed further in section 2.2.2. After the first officer’s query, the flight crew began to actively analyze the situation and take action, including looking for visible evidence of smoke or fire in the area behind the cockpit. The flight crew also proactively began troubleshooting to determine the source of the odor. Neither the UPS DC‑8 AOM nor the DC-8 Airplane Flight Manual states what procedures should be accomplished in the event that only fumes are present.

[ANALYSIS] The left engine fuel and hydraulic valves were found in the open or normal operating positions. Laboratory examination of the bulb filaments found that the left engine T-handle, which had been recovered in the stowed position, showed no evidence of stretching. These are not the positions to be expected for an engine that was shut down because of a fire indication. Based on the ATC transcript, the flightcrew did not talk about any attempts at engine restart(s). About 1 minute before impact, after the captain said, "We got an engine fire on the right side too, it 42 shows," the first officer asked the captain, "do we have any power?" The captain stated, "I'm not getting any response, man." It is possible that the first officer began the restart sequence on the secured left engine. Disassembly found that both power levers were in the full-forward position at impact, and an attempted restart on the left engine would have initially required that the left engine fire T-handle be pushed into the normal position. The result would have been to reopen the left engine fuel and hydraulic valves. The flightcrew may also have received a fire warning later in the flight as the fire continued to cause damage in the aft fuselage area. Their comment about "fire on the right side too, it shows" supports this possibility. There was no reasonable means for the flightcrew to observe fire in the aft fuselage from the cockpit. Consequently, their remarks about the location of the fire probably came from the cockpit engine T-handle fire warnings. Determination of the at-impact location of pointers and digits would have been valuable in determining what engine instrument information, if any, was available to the flightcrew immediately before impact, as well as whether or how electrical shorting and a fire in the aft fuselage affected cockpit instrumentation. However, laboratory examinations were not able to gain useful information from the instruments. Although the airplane appeared to be in a controlled, gradual, highspeed descent until just before it crashed, the tower recording of the pilots’ voices indicated that they were having difficulties controlling the airplane during the last portion of the flight, as well as in diagnosing mounting problems with the airplane. The airplane crossed the extended centerline of the runway, did not tum to final approach, and subsequently crashed in a nose-low left-wing-down attitude. It crashed nearly aligned with a wide residential avenue. Consequently, the Safety Board's analysis of this accident focused on the reasons for the in-flight fire and the reasons that the pilots were unable to perform a successful emergency landing. The analysis also examined the oversight of the maintenance and inspection of the airplane by the operator. 2.2 I The Fire The evidence found in the wreckage showed that the electrical power cables for the special mission equipment had not been installed in accordance with specifications. The improper installation left portions of the wires unprotected. by current limiters. The two large-diameter DC power wires that were retrieved from the wreckage showed evidence of arcing and they were "welded" together in the area }} ) 43 unprotected by the current limiters.

[FINDINGS] Their actions were particularly effective considering the number of persons in affected apartments, the substantial aircraft wreckage, and apartment building and vehicles fires. Command and control was well organized and expedited the dispatching of injured persons through the early establishment of an ambulance control center near the impact site. 3.1 49 3. CONCLUSIONS Findings 1. Weather was not a factor in the accident. 2. Air traffic services were proper and did not contribute to the causes of the accident. The pilots were properly trained and qualified for the flight. 4. The flightcrew experienced an in-flight fire leading to a request for an emergency landing. The special mission wiring was not installed properly, leading to a lack of overload current protection. 6. The FAA Form 337s provided instructions for the correct installation, and the mission power modifications made by another operator on 3 of the 18 special mission Learjets were correct. 7. Neither the mechanic(s) who installed the wiring nor the mechanic(s) holding the inspection authorization, who approved the installation, noted the nonconformity with the FAA Form 337 in the installation on N521PA and 14 other Learjets modified by the operator. 8. The in-flight fire most likely originated with a short of the special mission power supply wires in an area unprotected by current limiters. 9. The fire resulted in false engine fire warning indications to the pilots that led them to a shutdown of the left engine. 10. The intense fire, which burned through the aft engine support beam in flight, can be explained by a compromised fuel line resulting from a battery explosion. 11. 12. 13, 50 The in-flight fire caused substantial damage to the airplane. . structure and systems in the aft fuselage and may have precluded a successful emergency landing. At the time of impact, the left engine was not producing power; and the right engine was producing at least flight-idle power. The City of Fresno police, fire fighting, and rescue responses, which were assisted by units from Fresno Air Terminal, were timely and effective. 51 3.2 - Probable Cause The National Transportation Safety Board determines that the probable causes of this accident were: 1) improperly installed electrical wiring for special mission operations that led to an in-flight fire that caused airplane systems and structural damage and subsequent airplane control difficulties; 2) improper maintenance and inspection procedures followed by the operator; and, 3) inadequate oversight and approval of the maintenance and inspection practice by the operator in the installation of the special mission systems. 52 4. RECOMMENDATIONS © As a- result of the investigation of this accident, the National Transportation Safety Board makes the following recommendations: --to the Federal Aviation Administration: Publish an FAA Special Airworthiness Information Bulletin that describes the circumstances of this accident, including the consequences of improper installation of the special mission wiring, where electrical power wires were unprotected by current limiters.

[ANALYSIS] Battery explosion, specifically the left battery, could have compromised a fuel line. A fracture at the motive flow valve could have resulted in a high pressure (300 psi) spray that became ignited and impinged upon the aft engine support beam. The fact that the inlet fitting of the left motive flow valve was 44 gone and the threads inside the valve were undamaged is consistent with a leak at that location. It is also possible that the alcohol tank pressurization tube was melted through by the fire early in the sequence. The tube was not found in the wreckage; there was extensive fire damage in the aft fuselage area, and many nearby components were consumed or found partially melted. However, any compromise of the tube, early in fire propagation in the tailcone area, would have resulted in a ready source of engine bleed air into this enclosed space. There was evidence of soot in the aft tail section, specifically around lightening holes between the tail cone and vertical stabilizer; however, there was no evidence of severe smoke exit points from the fuselage. Further, most witnesses saw either no smoke or only light gray smoke trailing from the airplane before impact. Increased pressure flow from bleed air, impinging into the tailcone area, would have created a hotter fire, resulting in less smoke than expected emitting from the fuselage. 2.3 Airplane Handling The reason the flightcrew was unable to successfully land the airplane could not be conclusively determined. The 2 minutes of tower-recorded ‘intracockpit communications between the pilots prior to impact failed to provide sufficient data to determine the controllability of the airplane. The installation of a CVR and/or an FDR would have facilitated the determination of the events that led to the unsuccessful attempt to land the airplane. The Safety Board has addressed improved requirements for the installation and upgrading of CVR and FDR requirements for many facets of the aviation system. Although the installation of CVRs and FDRs are certainly important for passenger flight operations, they are also important for unique flight operations of aircraft with special mission equipment that operate over populated areas. This issue will be examined in the future by the Safety Board, and additional safety recommendations regarding operations, such as the accident flight, will be developed as appropriate. . . It is possible that the in-flight fire caused sufficient damage to the airplane structures and systems to render the airplane only partially controllable. -Although examination of the wreckage did not reveal a definitive reason for the loss of control, there is evidence that severe fire damage in the aft fuselage area occurred while the airplane was airborne. 45 Regarding engine power, the first officer asked the captain about 1 minute before impact, "do we have any power?” and the captain replied, "I'm not getting any response, man." However, engine disassembly evidence showed that although there was no indication of power on the left engine at impact, and it was only "windmilling,"” there was evidence of at least flight idle power on the right engine.

[CONCLUSIONS] They flew the airplane toward a right base for their requested runway, but the airplane continued past the airport. The flightcrew was heard on Fresno tower frequency attempting to diagnose the emergency conditions and control the airplane until it crashed, with landing gear down, on an avenue in Fresno. Both pilots were fatally injured. Twenty-one persons on the ground were injured, and 12 apartment units in 2 buildings were destroyed or substantially damaged by impact and fire. The National Transportation Safety Board determines that the probable causes of this accident were: 1) improperly installed electrical wiring for special mission operations that led to an in-flight fire that caused airplane systems and structural damage and subsequent airplane control difficulties; 2) improper maintenance and inspection procedures followed by the operator; and, 3) inadequate oversight and approval of the maintenance and inspection practice by the operator in the installation of the special mission systems. Safety issues in this report focused on maintenance, inspection and quality assurance. Safety recommendations concerning these issues were made to the Federal Aviation Administration, Phoenix Air, and the Department of Defense. NATIONAL TRANSPORTATION SAFETY BOARD WASHINGTON, D.C. 20594 AIRCRAFT ACCIDENT REPORT CRASH DURING EMERGENCY LANDING PHOENIX AIR, LEARJET 35A, N521PA FRESNO, CALIFORNIA DECEMBER 14, 1994: 1, FACTUAL INFORMATION 1.1 History of Flight On December 14, 1994, about 1146:23 pacific standard time (PST),! a Phoenix Air Group, Inc. (Phoenix Air) Learjet 35A, registration N521PA, crashed in Fresno, California. Operating under the call sign Dart 21, the flightcrew had declared an emergency inbound to Fresno Air Terminal (FAT) due to engine fire indications. They flew the airplane toward a right base for their requested runway, but the airplane continued past the airport. The flightcrew was heard on Fresno tower frequency attempting to diagnose the emergency conditions and control the airplane until it crashed, with landing gear down, on an avenue in Fresno. Both pilots were fatally injured. Twenty-one persons on the ground were injured, and 12 apartment units in 2 buildings were destroyed or substantially damaged by impact and fire. N521PA was a public-use aircraft,> under contract to the U. S. Air Force (USAF) to provide training for Air National Guard (ANG) fighters. The airplane had been modified with electronic equipment to satisfy the mission requirements. The mission was flown with a composite instrument flight rules/visual flight rules (FR/VFR) flight plan, following IFR for the departure and approach and VFR in the operating area.

[ANALYSIS] The improper installation left portions of the wires unprotected. by current limiters. The two large-diameter DC power wires that were retrieved from the wreckage showed evidence of arcing and they were "welded" together in the area }} ) 43 unprotected by the current limiters. This evidence indicates shorting and unlimited current flow for an extended period of time while the airplane was airbome. The investigation revealed that the proper installation specified that the special mission wires were to be attached to fittings in the generator control panel, where they would have been protected from overload during shorts by current limiters. Instead, the wires had been routed directly from the battery charging bus to 150 amp limiters before being routed to the control relay. Segments of the wires were unprotected by current limiters for about 18 inches, and they were found fused together between the battery bus and the current limiters. As a result of the wiring arrangement, large amounts of current could travel to the ground. 3 Fire damage to the wire insulation precluded a determination of the reasons for the shorting of the wires; however, one of these wires may have been frayed, and a short to ground (probably an airplane frame) occurred. Nevertheless, the evidence is conclusive that while the airplane was airborne, a severe short occurred on the left side of the aft fuselage in the electronics bay. Without the protection of the current limiters, the short was not interrupted. The evidence suggests that the arcing probably ignited wiring insulation and other combustible materials on the left side of the electronics bay. This caused damage to adjacent components. However, melting of the aft engine beam, the loss of Teflon electrical insulation on the engine fuel computer harness, and holes burned in the steel shield on the cabin conditioning hose required an intense fire directed at these items that was farther to the left side of the airplane. A diffuse fire limited to the electrical insulation and other solid combustibles in this area is unlikely to cause the type of fire damage noted. A "torch-like" flame from a pressurized fuel leak would be consistent with the fire damage noted on these items. The fact that these heavily fire-damaged components were in the same general location in the electronics bay of the airplane is also consistent with a burning fuel leak from a pressurized system. It is possible that the arcing or dead short drew excessive current, causing a battery to explode; this is supported by the conditions of the batteries and tiedowns. Two of the battery tiedown bolts were nearly straight, indicating the likelihood that they were not restraining substantial battery mass at the time of impact. Battery explosion, specifically the left battery, could have compromised a fuel line. A fracture at the motive flow valve could have resulted in a high pressure (300 psi) spray that became ignited and impinged upon the aft engine support beam. The fact that the inlet fitting of the left motive flow valve was 44 gone and the threads inside the valve were undamaged is consistent with a leak at that location.

[ANALYSIS] In response to the first officer’s urging, "pull up dude,” the captain replied about 35 seconds before impact, "I've got full right rudder in." In the absence of unknown flight contro! anomalies affecting directional control, this is consistent with asymmetrical thrust resulting from all or most of the power coming from the right engine. It is not known which cockpit indications, if any, the flightcrew had prior to impact to assist them in determining whether they had any engine power, or whether the indications were confusing due to electrical shorting and fire in the electronics bay in the aft fuselage. However, absent, incomplete, or misleading cockpit indications, combined with likely control malfunctions, specifically near full nosedown trim, led to an airplane condition that became increasingly difficult to control and difficult to diagnose. The damage to the aft engine support beam confirms that there was a severe fire directed onto this aluminum structure. The aluminum structure for the -beam was melted and missing for a considerable distance within the fuselage. It is possible that elevator and rudder control mechanisms were damaged by the fire and caused the loss of control. It is also possible that the loss of control related to the findings of the elevator trim about I degree from full nose down and the spoiler switch in the deployed position. Again, impact and postcrash fire damage precluded a determination of why the elevator trim was found near full nose down. Perhaps, the pilots activated the trim to compensate for other elevator problems, or the trim may have been activated by the fire in flight. There was no confirming physical evidence of spoiler deployment other than a guarded selector switch that was found in the deployed position. However, the possibility of deployed spoilers prior to impact is real because the pilots might have had reason to use them momentarily to slow the airplane for landing. If they 14The father of the first officer listened to the air traffic control recording of the flight (see letter, Appendix B) and identified the first officer and captain, respectively, as the pilots making the statements quoted in this sentence. This is the reverse attribution to that which was made on the transcript after mission pilots who had known the flightcrew listened to the tape. The father’s determination regarding the attributions for these two quotations appears correct. 46 did, there also could have been electrical anomalies caused by the in-flight fire that prevented spoiler retraction, or the pilots may have forgotten to retract them. However, even if the elevator trim and/or spoilers were misconfigured during the final phase of the emergency landing, the pilots should have been able to overcome the forces involved. Additionally, the observations of the various witnesses do not match a scenario in which the airplane is "nose heavy" and slowing.

[ANALYSIS] ANALYSIS General 0.0.0... ccescesssecessscecssecesseessssescsccesssccesseseesseccssesecosssnesssssasosuees 41 THE Fire «0... eeesseeseeceseeeseceeseeseseerenees cscsesecssasensescsrsnsesnseaceassnsoesenss 42 Alrplane Handling...............cssscscsscsssecssssssssccssscssesssssecesssssscsssssseoseees 44 Maintenance Oversight ............:ssccccccsesssssssesseeeees estesssnsesscsssesssscoccseees 46 Survival Facto! .........cssssccssssecessesssscsssecssssessssscsessssecssceesessssesssecoosees 48 CONCLUSIONS Findings ...........sccssssscccsessscessesecessssecesscesessnseecessnsccsssssesscesessnsessconeees 49 Probable Cause ............ssccsseccssccssssscsecesesscssssscessecssseasseseessssssseaseesooens 51 RECOMMENDATION G...............cssscssssssecsssreeeees deesacccesssceeesseeessees 52 APPENDIXES Appendix A--Investigation and Hearing...............ccsscccccccsssssserresevenees 55 Appendix B--Air Traffic Control Transcript.............sccsssscsssessessseeoes 56 Appendix C--FAA Forms 337 Regarding Special Mission Wiring Installation ...........00000 seseaeseaesesescseseseseseseceseseseseesessesceneseseseeeeesesenes 64 EXECUTIVE SUMMARY On December 14, 1994, about 1146:23 pacific standard time, a Phoenix Air Group, Inc. (Phoenix Air) Learjet 35A, registration NS21PA, crashed in Fresno, California. Operating under the call sign Dart 21, the flightcrew had declared an emergency inbound to Fresno Air Terminal due to engine fire indications. They flew the airplane toward a right base for their requested runway, but the airplane continued past the airport. The flightcrew was heard on Fresno tower frequency attempting to diagnose the emergency conditions and control the airplane until it crashed, with landing gear down, on an avenue in Fresno. Both pilots were fatally injured. Twenty-one persons on the ground were injured, and 12 apartment units in 2 buildings were destroyed or substantially damaged by impact and fire.

[ANALYSIS] Another example is an in-flight fire that originated in the left rear lavatory of an Air Canada DC-9 that forced the flightcrew to make an emergency landing on June 2, 1983, at the Greater Cincinnati Airport. In that fire, the first evidence of fuselage breakthrough occurred at the front of the aircraft, significantly forward of the initiation point. In addition, in a fire that originated below the floor immediately aft of the right galley on a Delta 737 while it was on the ground at Salt Lake City, Utah, on October 14, 1989, the first fuselage breach was in the first-class cabin, well forward of the fire’s origin. 81 Although 6R also contained a small amount of unburned combustible material (for example, newspaper wrapped around the industrial valves loaded in the outboard aft corner of the container), unlike the contents of containers in rows 8 and 9, the vast majority of easily burned unprotected material in 6R was consumed by the fire. 64 that container sustained fire damage only to the forward outboard post. Similarly, containers 9R and 8R contained significant amounts of unburned combustibles (such as paper items) after the fire. In analyzing the significance of the unburned materials in rows 8 and 9, the Safety Board recognized that when material towards the top of the containers in those rows burned, it could have formed an insulating layer of charred debris that would have slowed the downward progress of the fire.82 Postaccident examination revealed a significant amount of burned material (as well as unburned material) in that area. However, the high-pressure water streams from the firefighting efforts would have disrupted the arrangement of the burned and unburned materials, thus erasing any obvious signs of an insulating layer. Finally, investigators also noted that there was a "V" soot pattern on the interior of the fuselage originating at the floor level of the junction between 8L and 9L, but this is consistent with the exposure of that area when the fuselage separated and the fire vented at that point and is probably not related to the location of the fire’s origin. Thus, in comparing the fire damage in 6R with that in rows 8 and 9, it is possible that the fire in those rows was as significant as that in the area of 6R, but it might have started at or near the top of a container and was unable to progress very far into the volume of cargo loaded into those containers. In sum, there was insufficient reliable evidence to reach a conclusion as to where the fire originated. 2.3.2 Ignition Source of the Fire Because of an initial recognition among the fire investigators participating in the investigation that “V” burn patterns are generally highly significant, investigators examined the contents of 6R to try to identify a possible source of the fire.

[ANALYSIS] According to the Federal Register notice, the study was undertaken to compare the mission and requirements for civil airport fire services to those of the Department of Defense. On August 1, 1996, the Safety Board commented: [T]he current mission set forth in 14 CFR Part 139 to “provide an escape path from a burning airplane” no longer suffices. The Safety Board supports a full study of the mission statement by the FAA with a view towards providing adequate [ARFF] resources to rapidly extinguish aircraft interior fires and to extricate aircraft occupants from such interior fires. All aspects of this issue, including staffing, extinguishing agents, firefighter training, and response times, should be evaluated and compared with DOD standards to develop a broader mission statement that includes interior cabin fire suppression and extrication of aircraft occupants. 88 The assistant fire chief who served as the initial incident commander testified that at the flight engineer’s suggestion a telephone call was placed to the airplane manufacturer (Douglas) in an unsuccessful attempt to determine whether there were alternate means for entering the airplane. 89 Air Canada DC-9-32 in Covington, Kentucky, June 2, 1983 (23 persons killed by smoke/and or fire); USAir 737 collision with a Skywest Fairchild Metro 227 in Los Angeles, California, on February 1, 1991 (22 persons killed by smoke/and or fire); Northwest Airlines DC-9 collision with a Northwest 727 in Detroit, Michigan, on December 3, 1990 (eight persons killed by smoke and/or fire); Air Transport International DC-8-62 in Jamaica, New York, on March 12, 1991 (freight only); Ryan International Airlines B-727 in Hartford, Connecticut, on May 3, 1991 (freight only); and TWA Lockheed L1011 in Jamaica, New York, on July 30, 1992. 75 Accident history suggests that the environment inside a burning airplane’s interior may be beyond the current technological capabilities of fire departments to extinguish within adequate time frames to successfully evacuate occupants or protect cargo. The Safety Board is aware that the FAA has researched fire extinguishing systems for airplane interiors, including testing of a water spray system that would discharge water into a particular area of the airplane when triggered by sensors in that area. Because the system would discharge water only to a focused area of potential fire, it would minimize the total amount of water that would need to be carried on board, thereby reducing the weight penalty of such a system. FAA tests showed that when this system was used to fight a fire, it delayed the onset of flashover, reduced cabin air temperatures, improved visibility, and increased potential survival time. The Safety Board is concerned about the number of losses that have occurred and concludes that currently, inadequate means exist for extinguishing on-board aircraft fires.

[ANALYSIS] Although the investigation examined this as a possible ignition scenario, it could not be determined whether the chemicals in the synthesizer played any role in igniting the fire. The investigation could not develop a viable and convincing scenario to explain how the synthesizer could have started a fire. Further, although the cargo debris from all of the cargo containers that had been carrying general cargo was also examined for possible ignition sources, it was not initially examined in as much detail as the items known to have been loaded into 6R because of the volume and condition of the debris. When a more detailed examination of that cargo debris was conducted later in the investigation, no ignition sources were identified in that cargo debris. However, because of the deteriorated condition of the cargo debris and the possibility that some cargo had been completely destroyed in the fire, the Safety Board could not rule out that on the accident flight, an ignition source was present in one of those cargo containers. In light of the discovery of several shipments of marijuana on board the accident flight and the suggestion of one of the parties that marijuana is subject to spontaneous combustion, investigators considered this as a possible ignition source. The Safety Board recognizes that some organic matter, such as hay, can generate a biological/organic reaction producing heat and combustion if it is both wet and sufficiently compressed yet still exposed to oxygen. However, all of the marijuana shipments on board the airplane had been vacuum packed. (The police investigator who documented the marijuana seizures testified that the specially trained dog would have detected the presence of additional packages of marijuana, if any had been on board, even if the packages were completely consumed in the fire.) The police investigator who documented the marijuana seizures explained that shippers of contraband such as marijuana attempt to reduce the size of the package by “using a vacuum to vacuum out all the air and get it as compact as possible.” Thus, although the marijuana would have been compressed, there would have been little or no oxygen available to permit or support the biological reaction needed to lead to spontaneous combustion. Further, neither the police investigator nor any of the fire experts or consultants questioned during the course of the Safety Board’s investigation were aware of a fire being initiated by spontaneous combustion of a marijuana shipment. Therefore, spontaneous combustion of a marijuana shipment was ruled out as a possible ignition source. 66 Finally, all of the airplane systems were examined for possible ignition sources; the electrical system showed no evidence of arcing, and none of the other aircraft systems showed any evidence of malfunction. Further, neither the crew interviews nor the FDR or CVR data indicated any failures or malfunctions in the airplane’s systems that might have played a part in initiating the fire. Therefore, aircraft systems were ruled out as a possible ignition source for the fire.

[ANALYSIS] Container 9R’s aluminum roof and two of its three Lexan walls were consumed by fire, as was the upper portion of the third Lexan wall; its inboard cornerposts were intact, but its outboard forward and aft cornerposts measured 1 foot 3.5 inches and 1 foot 5 inches, respectively. However, 9L contained a significant quantity of undamaged materials81 with a low melting point (polyurethane, polystyrene, and polyethylene), and the cornerposts of 79 It should be noted that a FedEx mechanic who was standing near the cargo door indicated that at about the same time that flames were first seen venting from the aft section of the airplane in the 8/9 area, or possibly earlier, flames were also starting to break through the side of the fuselage in the area of 6L. His observations were not contradicted by the other eyewitnesses. Rather, they suggest that flames might have been breaking through the fuselage at more than one place at the same time. The video footage showing flames breaking through farther aft was taken from the right of the airplane and would not have captured a breakthrough of flames on the left side. 80 The breakthrough principle is illustrated by several previous main cabin aircraft fires. For example, a cabin fire occurred in Atlanta in 1995 on a ValuJet DC-9 after an engine failure started in the aft end of the cabin from a failed engine compressor disk that punctured a fuel line and penetrated the cabin. (See, National Transportation Safety Board. 1996. Uncontained Engine Failure/Fire, ValuJet Airlines Flight 597, DC-9-32, N908VJ, Atlanta, Georgia, June 8, 1995. NTSB/AAR-96/03. Washington, DC.) In a report submitted to the Safety Board, an FAA fire specialist noted that in that accident, in spite of rapid fire department response, the fire gutted the cabin and penetrated the fuselage skin just behind the cockpit—at the opposite end of the cabin from where the fire started. Another example is an in-flight fire that originated in the left rear lavatory of an Air Canada DC-9 that forced the flightcrew to make an emergency landing on June 2, 1983, at the Greater Cincinnati Airport. In that fire, the first evidence of fuselage breakthrough occurred at the front of the aircraft, significantly forward of the initiation point.

[ANALYSIS] One factor that investigators considered was the "V" burn pattern that originated at container 6R. It is a basic premise of fire science that such a “V” pattern often points to the origin of a fire. However, as explained in the National Fire Protection Association’s Guide for Fire and Explosion Investigations, NFPA 921, “each time another fuel package is ignited or the ventilation to the fire changes, the rate of energy production and heat distribution will change. Any burning item can produce a plume and, thus a ‘V’ pattern. Determining which pattern was produced at the point of origin by the first material ignited becomes more and more difficult as the size and duration of the fire increases.” [Par. 3-7]77 Several areas in the main cabin cargo compartment exhibited extensive fire damage; however, the deepest and most severe heat and fire damage was found in and around container 6R.78 More of 6R's structure was consumed than of any other container, and it was the only container that exhibited severe floor damage, which was likely caused by burning/melting Lexan wall material that fell inward onto the unprotected container floor. (Container 6R was one of the more sparsely loaded containers, enabling fire to reach the floor level without having to consume much cargo in the process.) Further, 6R was the only container to exhibit heat damage on its bottom surface, and the area below container 6R showed the most extensive evidence of scorching of the composite flooring material. In addition, the overall burn damage pattern to the cargo containers in the main cargo cabin showed that the deepest burned-out area centered over container 6R and that the cargo containers surrounding 6R and the contents in these containers were all burned to a greater depth along the sides common to container 6R. There was heat damage to the cabin floor just aft of container 9L, but there was no cargo container in this area on the accident flight. This damage to the floor was consistent with burning material falling from the burning fuselage crown or contents from cargo container 9L falling into this empty space. Two aerosol cans found in this area showed evidence that they had been heated by fire on the outside of the cans, and that they had overpressurized and ruptured. If the fire had not burned so long, the "V" burn damage pattern and the extensiveness of the fire damage to 6R would have been stronger evidence of a fire originating inside 6R. Further, the deep burn and severe damage found in container 6R could also be accounted for by the fact that it was relatively empty and therefore largely unprotected by cargo. Thus, the Lexan side walls and nylon curtain could have fallen directly onto the floor of 6R and burned there, resulting in the severe damage to the floor of 6R and the exterior surfaces of the synthesizer.

[ANALYSIS] Evidence from crew duty time, flight time, rest time, and off-duty activity patterns did not indicate that behavioral or psychological factors related to fatigue affected the flightcrew on the day of the accident. The smoke detection system installed on the airplane functioned as intended and provided the crewmembers with sufficient advance warning of the in-flight fire to enable them to land the airplane safely. The ATC personnel involved with the flight were all properly certificated and qualified. The Boston Center ARTCC and New York TRACON controllers responded appropriately once they were aware of the emergency and provided appropriate and needed information to assist the crew in the emergency descent and landing. The airplane was properly certificated, equipped, and maintained in accordance with applicable regulations. No evidence of systems, mechanical, or structural failures was found. 2.2 Flightcrew Performance 2.2.1 Crew Coordination Although the airplane was landed successfully, several required items were not accomplished during the descent and landing. The flight engineer failed to perform step No. 6 of the “Cabin Cargo Smoke Light Illuminated” checklist (pulling the cabin air shutoff T-handle).71 If he had done so, airflow would have been shut off to the main cargo deck area while being maintained to the cockpit. The Safety Board concludes that the flight engineer’s failure to pull the cabin air shutoff T-handle, as required by the “Cabin Cargo Smoke Light Illuminated” checklist, allowed the normal circulation of air to continue to enter the main cargo area, thereby providing the fire with a continuing source of oxygen and contributing to its rapid growth. However, the Safety Board could not determine the degree to which it might have contributed to the severity of the fire. The flight engineer also failed to complete step No. 7 of the “Cabin Cargo Smoke Light Illuminated” checklist (to maintain a 0.5 psi differential cabin pressure). As a result, the occupants were unable to immediately open and exit from the primary evacuation exits (the L1 71 Although the CVR recorded the flight engineer stating, “pull cabin air” at 0538:40, its position after the accident indicates that the cabin air shutoff T-handle had not been pulled. 58 and R1 doors) because the airplane was still pressurized. The flight engineer acknowledged that instead of manually maintaining the appropriate pressure differential, after he had placed the outflow valve control in the manual position, he only “cranked it open a couple of times [turns].” Because they were at 33,000 feet and operating on only one pressurization pack, the outflow valve would have been almost completely closed before the flight engineer cranked it. As demonstrated in the Safety Board’s test on a similar DC-10, manually cranking the outflow valve control two times will not perceptibly open the outflow valve from fully closed on a static airplane. The Safety Board concludes that the evacuation was delayed because the flightcrew failed to ensure that the airplane was properly depressurized.

[CONCLUSIONS > FINDINGS] No evidence of systems, mechanical, or structural failures was found. 5. The flight engineer’s failure to pull the cabin air shutoff T-handle, as required by the “Cabin Cargo Smoke Light Illuminated” checklist, allowed the normal circulation of air to continue to enter the main cargo area, thereby providing the fire with a continuing source of oxygen and contributing to its rapid growth. However, the Safety Board could not determine the degree to which it might have contributed to the severity of the fire. 6. The evacuation was delayed because the flightcrew failed to ensure that the airplane was properly depressurized. 7. The captain did not adequately manage his crew resources when he failed to call for checklists or to monitor and facilitate the accomplishment of required checklist items. 8. Crewmembers who do not use protective breathing equipment during a smoke or fire emergency may place themselves at unnecessary risk in attempting to address or escape from the situation. 9. Crewmembers may not be adequately aware that attempting to open a passenger exit door when the airplane is still pressurized may result in the door not opening. 77 10. The DNA synthesizer was not completely purged of volatile chemicals (including acetonitrile and tetrahydrofuran) before it was transported on board flight 1406. 11. The presence of the aerosol cans, the containers of acidic liquid, as well as several packages of marijuana on board the accident flight illustrates that common carriers can be unaware of the true content of many of the packages they carry. 12. The transportation of undeclared hazardous materials on airplanes remains a significant problem and more aggressive measures to address it are needed. 13. The Department of Transportation hazardous materials regulations do not adequately address the need for hazardous materials information on file at a carrier to be quickly retrievable in a format useful to emergency responders. 14. FedEx’s policy of providing information only to the Safety Board after the Safety Board initiates an investigation is inconsistent with the need to quickly provide emergency responders with essential information to assess the threat to themselves and the local community. 15. More effective preparation for emergencies involving hazardous materials and a system for coordination among the Air National Guard, Stewart International Airport management, and all local and State emergency response agencies are needed. 16. Airport emergency plans should specifically address hazardous materials emergencies. 17. Currently, inadequate means exist for extinguishing on-board aircraft fires. 18. In addition to the safety benefits provided by on-board extinguishing systems, aircraft rescue and firefighting capabilities must also be improved so that firefighters are able to extinguish aircraft interior fires in a more timely and effective manner. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was an in-flight cargo fire of undetermined origin. 78

[ANALYSIS] Further, the deep burn and severe damage found in container 6R could also be accounted for by the fact that it was relatively empty and therefore largely unprotected by cargo. Thus, the Lexan side walls and nylon curtain could have fallen directly onto the floor of 6R and burned there, resulting in the severe damage to the floor of 6R and the exterior surfaces of the synthesizer. When Lexan is heated, it typically burns, melts, and puddles, producing heat that would be sufficient to cause the damage to container 6R and its contents. Thus, a fire that 77 See also Par. 3-7.2, which states, “[a]reas of great damage are indicators of a high heat release rate, ventilation effects, or long exposure. Such areas, however, are not always the point of fire origin.” 78 In every other cargo container there was a layer of unburned cargo covering the container floor. 63 originated outside of 6R but eventually spread to that area could have resulted in a similar damage pattern. Comments on the CVR suggest that the smoke detector activation sequence might have begun with detector number 9 and initially moved forward; this suggests that the fire might have started aft of row 6. Further, some of the first flames to have breached the crown were observed approximately above the area occupied by cargo container rows 8 and 9.79 Although the smoke detector activation sequence and location of the early breakthrough of flames cannot be considered reliable indicators of a fire’s initial location,80 a possible connection between these factors and the location of the fire’s origin could not be discounted. Therefore, the Safety Board also considered the possibility that the fire originated aft of container row 6. Although there was some significant burn damage to the containers in rows 8 and 9, their contents, and the surrounding area of the aircraft, this damage appeared to have been less than the damage in the area of container 6R. For example, container 8L’s aluminum roof and two of its three Lexan walls were consumed by fire, as was the upper two-thirds of the third Lexan wall; its inboard forward and aft cornerposts measured 5 feet 10 inches and 5 feet 2 inches, respectively, and its outboard forward and aft cornerposts measured 1 foot 4.5 inches and 3 feet 6.5 inches, respectively. Container 9R’s aluminum roof and two of its three Lexan walls were consumed by fire, as was the upper portion of the third Lexan wall; its inboard cornerposts were intact, but its outboard forward and aft cornerposts measured 1 foot 3.5 inches and 1 foot 5 inches, respectively.

[ANALYSIS] As discussed in section 1.6.4, when there is no electric power to the airplane the motor that operates the door is powered by a charged air bottle. If an attempt is made to open the door when the cabin pressure differential is above 0.5 psi, the bottle pressure will bleed off and the door will not open. Although the lack of the L-1 door as an escape route was not a significant factor in this accident, the Safety Board is concerned that under other circumstances the loss of a passenger exit door could have serious safety consequences. The Safety Board concludes that crewmembers may not be adequately aware that attempting to open a passenger exit door when the airplane is still pressurized may result in the door not opening. Therefore, the Safety Board believes that the FAA should require all Part 121 operators of airplanes that rely on air pressure to operate exit doors to make crewmembers aware of the circumstances of this accident and remind them of the need to ensure that the airplane is depressurized before attempting to open the passenger exit doors in an emergency. 2.3 Fire Initiation 2.3.1 Location from Which the Fire Might Have Initiated Because the fire burned for about 4 hours after smoke was first detected in the cabin cargo compartment, under changing conditions, much of the potentially helpful evidence was destroyed by the fire.76 The growth of the fire was likely affected by the failure to pull the cabin air shutoff T-handle and the opening of doors L1 and R1 about 0556, which (even though L1 did not open completely) would have provided additional ventilation (oxygen), and thus increased the rate of fire growth. An even greater addition of oxygen occurred about 0650 when the cargo door was opened. (Witnesses reported that flames first broke through the fuselage shortly after the cargo door was opened. The airport operations log recorded that flames breached the crown of the fuselage about 0655, which was 1 hour after the airplane landed and about 1 hour and 19 minutes after the illuminated smoke detector lights were first noted by the flightcrew.) Despite the length of time the fire burned and the resulting destruction of potentially 76 It should nonetheless be noted that even after the prolonged fire, many containers (including those in rows 1, 2, 3, 14, 15, and 16) were not severely compromised, or still contained substantial quantities of highly combustible materials (such as magazines, technical manuals, dry cleaning bags, bubble wrap, and clothing) that remained unburned. 62 helpful evidence, the postfire condition of the airplane and its contents were nonetheless examined for clues as to the location from which the fire might have initiated. One factor that investigators considered was the "V" burn pattern that originated at container 6R. It is a basic premise of fire science that such a “V” pattern often points to the origin of a fire.

[ANALYSIS] In sum, there was insufficient reliable evidence to reach a conclusion as to where the fire originated. 2.3.2 Ignition Source of the Fire Because of an initial recognition among the fire investigators participating in the investigation that “V” burn patterns are generally highly significant, investigators examined the contents of 6R to try to identify a possible source of the fire. All items found in, or known to have been shipped in, container 6R were examined in detail (see section 1.14.2.1.1). In particular, the Safety Board examined the DNA synthesizer as a potential source of ignition because of the chemical smell noticed inside the unit and because the other items in that container were ruled not likely to have provided a source of ignition. The nature and degree of the fire damage to the synthesizer, particularly the heavy damage to the internal circuitry of the synthesizer’s controller panel (which is made of a low-flammability material), was thought to be suggestive of a source of fuel inside the synthesizer. (However, given the presence of volatile chemicals, and possibly vapors, inside the synthesizer, it is possible that a fire ignited external to the synthesizer could have produced a damage pattern similar to that resulting from an internal ignition source.) 82 According to the National Fire Protection Association’s Guide for Fire and Explosion Investigations, NFPA 921, “[a] protected area results from an object preventing the products of combustion from depositing on the material that the object protects, or prevents the protected material from burning.” [Par. 4-15.2] 65 Tests of the liquids from the accident synthesizer showed that flammable chemicals (THF and acetonitrile) were still present in the bottles on the machine after the fire. The quantity of chemicals remaining in the synthesizer’s bottles after the fire was insufficient to have caused the extensive internal fire damage to the synthesizer and the cargo container. However, it is likely that significant amounts of the chemicals were consumed in the prolonged and intense fire and thus the synthesizer probably contained much larger quantities of these flammable chemicals before the fire. (The presence of firefighting agent inside most of the bottles and the damage to many of the tubes that entered the bottles indicates that the bottles had been open to the atmosphere during at least part of the fire sequence.) These volatile chemicals—particularly the THF—could ignite a fire. THF, which is highly flammable under any circumstances, can also form unstable peroxides that can explode on contact with certain other materials or autoignite (spontaneously explode) in sufficient concentrations. Although the investigation examined this as a possible ignition scenario, it could not be determined whether the chemicals in the synthesizer played any role in igniting the fire. The investigation could not develop a viable and convincing scenario to explain how the synthesizer could have started a fire.

[ANALYSIS] FAA tests showed that when this system was used to fight a fire, it delayed the onset of flashover, reduced cabin air temperatures, improved visibility, and increased potential survival time. The Safety Board is concerned about the number of losses that have occurred and concludes that currently, inadequate means exist for extinguishing on-board aircraft fires. Therefore, the Safety Board believes that the FAA should reexamine the feasibility of on-board airplane cabin interior fire extinguishing systems for airplanes operating under 14 CFR Part 121 and, if found feasible, require the use of such systems. The Safety Board realizes that requiring on-board extinguishing systems may not entirely resolve these safety concerns because they may become disabled by crash impacts. Further, the Safety Board realizes that the full implementation of such technology will require a number of years. Therefore, the Safety Board concludes that in addition to the safety benefits provided by on-board extinguishing systems, ARFF capabilities must also be improved so that firefighters are able to extinguish aircraft interior fires in a more timely and effective manner. Therefore, the Safety Board believes that the FAA should review the aircraft cabin interior firefighting policies, tactics, and procedures currently in use, and take action to develop and implement improvements in firefighter training and equipment to enable firefighters to extinguish aircraft interior fires more rapidly. 76 3. CONCLUSIONS 3.1 Findings 1. The flightcrew was properly certificated and qualified in accordance with the applicable regulations and company requirements. Evidence from crew duty time, flight time, rest time, and off-duty activity patterns did not indicate that behavioral or psychological factors related to fatigue affected the flightcrew on the day of the accident. 2. The smoke detection system installed on the airplane functioned as intended and provided the crewmembers with sufficient advance warning of the inflight fire to enable them to land the airplane safely. 3. The Boston Center air route traffic control center and New York terminal radar approach control controllers responded appropriately once they were aware of the emergency and provided appropriate and needed information to assist the crew in the emergency descent and landing. 4. The airplane was properly certificated, equipped, and maintained in accordance with applicable regulations. No evidence of systems, mechanical, or structural failures was found. 5. The flight engineer’s failure to pull the cabin air shutoff T-handle, as required by the “Cabin Cargo Smoke Light Illuminated” checklist, allowed the normal circulation of air to continue to enter the main cargo area, thereby providing the fire with a continuing source of oxygen and contributing to its rapid growth.

[ANALYSIS] The Safety Board recognizes that after this accident Stewart revised its emergency plan, and that airport operations personnel at Stewart have acknowledged the need to address those deficiencies in the airport’s emergency plan. However, the Safety Board is concerned that FAA requirements87 do not specifically address the need to prepare for hazardous materials emergencies, and that other airports may be similarly unprepared for hazardous materials emergencies. The Safety Board concludes that airport emergency plans should specifically address hazardous materials emergencies. Therefore, the Safety Board believes that the FAA should require all certificated airports to coordinate with appropriate fire departments, and all State and local agencies that might become involved in responding to an aviation accident involving hazardous materials, to develop and implement a hazardous materials response plan for the airport that specifies the responsibility of each participating local, regional, and State agency, and addresses the dissemination of information about the hazardous materials involved. Such plans should take into consideration the types of hazardous materials incidents that could occur at the airport based on the potential types and sources of hazardous materials passing through the airport. The Safety Board also believes that the FAA should require airports to coordinate the scheduling of joint exercises to test these hazardous materials emergency plans. 86 The assistant fire chief on duty served as incident commander until 0700, when the fire chief arrived on scene. 87 14 CFR 139.325 specifies what must be included in airport emergency plans of airports certificated under Part 139. 74 Firefighters were positioned on scene before the airplane landed and began firefighting efforts immediately. Although the firefighters initially attempted to conduct an interior attack on the fire from the foyer area, the location of the cargo containers prevented them from approaching the seat of the fire. After the cargo door was opened, firefighters observed orange flames and heavy smoke in the airplane, and the incident commander evacuated them from the airplane. The initial incident commander’s decision to evacuate the firefighters from the interior of the airplane was appropriate given the danger posed by the smoke and fire-filled airplane. However, the initial incident commander acknowledged that use of the SPAAT tool to penetrate the fuselage was delayed while he attempted to accommodate the flight engineer’s request that damage to the airplane be minimized.88 Although it is not clear whether an earlier entry would have improved the effectiveness of the firefighting efforts in this case, the Safety Board is concerned that more aggressive measures to enter the airplane, such as use of a fuselage penetrating tool, were not taken sooner. The Safety Board notes that the ANG fire chief testified that based on “lessons learned” from this accident, if a similar situation were to occur, he would immediately “get right in there with a hand line and deploy some type of penetrating tool on the outer skin of the aircraft.” The Safety Board has long been concerned about the lack of success of airport fire departments in extinguishing interior fires.89 On June 4, 1996, the FAA published “Airport Rescue and Firefighting Mission Response Study,” in the Federal Register and invited comments from interested parties. According to the Federal Register notice, the study was undertaken to compare the mission and requirements for civil airport fire services to those of the Department of Defense.

[ANALYSIS] No airplane debris (except for a main landing gear door) was found before the berm, indicating the airplane structure was intact before this point. Ground scars from the left main and nose landing gears were found about 620 feet from the end of the runway. The airplane then became airborne again, and the cockpit and the left wing impacted an initial group of trees located about 900 feet from the end of the runway. This impact caused the cockpit to break open and the left wing fuel tank to rupture, allowing a fuel-air mixture to ignite. The airplane then impacted the ground about 1,250 feet from the end of the runway, as evidenced by ground scars from the entire lower surface of the airplane. The airplane then slid 400 feet and struck two large diameter trees. The impacts breached the passenger cabin, separating it into two sections and allowing a large amount of fuel, fuel vapor, and fire to enter the cabin. The fuselage traveled another 150 feet before coming to a stop. The airplane structure continued to burn, and the fire eventually consumed the entire fuselage and cabin interior. The simultaneous impacts with two large diameter trees caused numerous blunt force fatal injuries to passengers. One of these tree strikes occurred on the left side of the fuselage several rows aft of the main cabin door.209 The other tree strike occurred in the area where the left inboard wing attached to the fuselage. This tree breached the left wing fuel tank and the cabin in the area of the left overwing exit, cut through the cabin as the airplane continued to slide, and exited the right side of the aft cabin. Figure 8 shows the approximate tree strike areas in the cabin. 209  This tree was uprooted and carried with the airplane to its final location. Analysis National Transportation Safety Board A I R C R A F T Accident Report 87 Approximate Tree Strike Areas and Injury Information Figure 8.  Note: On the basis of body location information and autopsy results, the Safety Board determined that, before the attempted takeoff, the passengers assigned to seats 2A and 2B most likely switched seats, the passenger assigned to seat 10D most likely moved to seat 7B, and the nonrevenue passenger most likely took seat 12B. Note: Seat positions showing more than one color indicate more than one reported major pathologic finding in the autopsy reports. Analysis National Transportation Safety Board A I R C R A F T Accident Report 88 The captain’s body was found in the forward area of the passenger cabin. The seriousness of the captain’s blunt force injuries would have precluded him from initiating any postaccident movement. Thus, the captain (and his seat) must have been separated from the cockpit during the impact sequence. The first officer was found in the remnants of his seat secured by the seat’s restraint system. He was extricated from the cockpit wreckage by first responders to the accident scene and transported to the hospital, as discussed in section 2.3.2. The flight attendant’s body was found close to his jumpseat.

[ANALYSIS] Thus, the captain (and his seat) must have been separated from the cockpit during the impact sequence. The first officer was found in the remnants of his seat secured by the seat’s restraint system. He was extricated from the cockpit wreckage by first responders to the accident scene and transported to the hospital, as discussed in section 2.3.2. The flight attendant’s body was found close to his jumpseat. His autopsy report and those of several passengers in the forward area of the cabin showed a relative lack of blunt force injuries and evidence of smoke inhalation. These findings indicated that these occupants survived the impact but succumbed to the effects of the postcrash fire and smoke. It was not possible to determine the length of time that these occupants survived, but it is important to note that all of these occupants were found close to their seats. Several passengers who were seated in the forward left part of the cabin were found near each other in the aft right part of the cabin. It is likely that the forward tree strike caused these passengers and their seats to be displaced. Most of these passengers sustained fatal blunt force injuries. Most of the passengers seated in rows 7 through 9 (the overwing exit row and the rows immediately in front of and in back of it) were killed immediately by the flash fire that occurred after the left wing fuel tank was breached by the aft tree strike and fuel, fuel vapor, and fire entered the cabin. Two passengers were found outside of the cabin on the left side of the fuselage. These passengers sustained fatal thermal injuries, and neither sustained serious blunt force injuries. Evidence indicated that the passengers were likely thrown outside (through the break in the fuselage caused by the aft tree strike) when the airplane came to a stop. Several passengers who were seated in the aft cabin sustained some blunt force injuries, and most showed evidence of smoke inhalation. It was not possible to determine the length of time that these passengers survived, but it is important to note that all of the passengers were found close to their seats. Emergency Response 2.3.2  According to the aircraft performance study for this accident, the airplane struck trees and terrain west of the airport about 0606:35. The sounds of the accident took about 5 seconds to reach the tower. The controller stated that, after hearing these sounds, he saw a fire west of the airport. The ATC transcript showed that the controller activated the crash phone about 0607:17. Thus, about 37 seconds had elapsed between the time that the accident sounds could be heard in the tower and the time that the controller activated the crash phone. During this time, the controller had to turn around and look outside the tower cab window; assess the situation; and recognize that, even though the airplane Analysis National Transportation Safety Board A I R C R A F T Accident Report 89 had been cleared to take off from runway 22, the airplane was actually located off the departure end of runway 26. The ATC transcript showed that the airport fire department responded to the crash phone about 0607:22.

[ANALYSIS] Analysis National Transportation Safety Board A I R C R A F T Accident Report 89 had been cleared to take off from runway 22, the airplane was actually located off the departure end of runway 26. The ATC transcript showed that the airport fire department responded to the crash phone about 0607:22. According to the ATC transcript, the controller announced an “alert three” and indicated that a Comair jet was located at the west side of the airport just off the approach end of runway 8. By providing clear and accurate information about the airplane’s status and location, the controller’s actions complied with the LEX ATCT’s standard operating procedures and letter of agreement with the airport board regarding emergency notifications. The first emergency responders to arrive on scene were a LEX public safety officer and a police officer from the city of Lexington. They arrived (independently of each other) in the general vicinity of the accident about 5 1/2 minutes after receiving notification of the accident and reached the fuselage within 3 1/2 minutes despite restricted visibility caused by tall vegetation near the accident site. The first ARFF vehicle arrived on scene about 11 minutes after alert 3 notification using route information provided by another LEX public safety officer (who was in the general vicinity of the accident site but was not yet on scene). This route was the most effective direct route available to the accident site. This ARFF vehicle immediately began fire suppression and knocked down most of the fire. The second ARFF vehicle arrived on scene shortly afterward and began additional fire suppression. The LEX operations center incident report showed that the fire was controlled about 3 minutes after the first ARFF vehicle arrived on scene. The assistant ARFF chief (who arrived on scene in the first ARFF vehicle) stated that, upon reaching the airplane, its top was gone, and its sides were mostly gone. Other first responders reported that the cabin interior was completely involved (that is, completely on fire) at the time of their arrival on scene. The LEX public safety officer and the police officer from the city of Lexington were able to free the first officer from the cockpit wreckage. Another LEX public safety officer and the city of Lexington police officer transported the first officer to the hospital in a sport-utility vehicle instead of waiting for an ambulance. This LEX public safety officer estimated that they arrived at the hospital (located about 7 miles from the accident site) before 0630. Because of the serious traumatic injuries that the first officer sustained from the accident, it was imperative that he be quickly transported to and receive immediate treatment from a trauma center. The Safety Board concludes that the first officer’s survival was directly attributable to the prompt arrival of the first responders; their ability to extricate him from the cockpit wreckage; and his rapid transport to the hospital, where he received immediate treatment. The Safety Board further concludes that the emergency response for this accident was timely and well coordinated.

[ANALYSIS] The Safety Board concludes that the first officer’s survival was directly attributable to the prompt arrival of the first responders; their ability to extricate him from the cockpit wreckage; and his rapid transport to the hospital, where he received immediate treatment. The Safety Board further concludes that the emergency response for this accident was timely and well coordinated. Analysis National Transportation Safety Board A I R C R A F T Accident Report 90 Summary of Survival Factors 2.3.3  The Safety Board’s definition of a survivable accident is as follows:210 An accident in which the forces transmitted to the occupant(s) through the seat and restraint system do not exceed the limits of human tolerance to abrupt accelerations and in which the structure in the occupants’ immediate environment remains substantially intact to the extent that a livable volume is provided for the occupants through the crash sequence. The captain and the passengers who received fatal blunt force injuries as a result of the simultaneous tree strikes were clearly in areas of the airplane in which the forces transmitted to the occupants exceeded the limits of human tolerance. Most of the passengers in the overwing area of the cabin did not experience similar forces; however, the large amount of fuel, fuel vapor, and fire forced into the cabin by the aft tree strike made the cabin environment immediately unsurvivable for those passengers. The flight attendant and most of the passengers in the forward and aft areas of the cabin also did not experience forces that exceeded the limits of human tolerance and maintained a livable volume of occupiable space for an undetermined amount of time. As stated in section 2.3.1, the flight attendant and these passengers were found close to their assigned seats. Because the impact forces did not exceed the limits of human tolerance and occupiable space was maintained for some of the airplane occupants, this accident was partially survivable. However, the environment inside the airplane deteriorated quickly as a result of the postcrash fire and smoke, which did not allow sufficient time or means for those occupants to evacuate. Efforts to Mitigate Airport Surface Operation Errors 2.4  Surface operation errors, including those that lead to wrong runway takeoff events, can be mitigated in several ways, such as improved flight deck procedures, the implementation of cockpit moving map displays or cockpit runway alerting systems, improved airport surface marking standards, and ATC policy changes. These systemwide interventions, which are discussed in sections 2.4.1 through 2.4.5, can provide the necessary redundancy to reduce the opportunity for human error during surface operations and, if an error were to occur, to stop it before it becomes catastrophic. These interventions can also help prevent runway incursions, which is an issue on the Safety Board’s list of Most Wanted Transportation Safety Improvements. 210  This definition was cited in the Safety Board’s 1981 special study on cabin safety in large transport aircraft.

[CONCLUSIONS] CONCLUSIONS Findings Probable Cause RECOMMENDATIONS APPENDIXES Appendix A--investigation and Hearing Appendix B--Personnel information Appendix C--Airplane information Appendix D--DHC-8 Engine Fire (In flight) Checklist Appendix E--Cockpit Voice Recorder Transcript EXECUTIVE SUMMARY On April 15, 1988, Korizon Air, inc., flight 2658, a 37-passenger deHavilland DHC-8 registered in the United States as N819PH, was a regularly scheduled passenger-carrying flight between Seattle, Washington, and Spokane, Washington. Shortly after takeoff, with ihe captain at the controls, the aircrew noted a power loss on the right engine. The captain made the decision to return to Seattle for a precautionary ianding. After lowering the landing gear on final approach, a massive fire broke out in the right engine nacelle. After the first officer shut down the engine, the captain proceeded to land the airplane; however, shortly after touchdown, the crew realized that almost all directional control and braking capability was lost. The airplane departed the paved surface of the runway, crossed a grass median area, entered the paved rarp area, and struck a runway designator sign, several baggage carts, and two jetways. The airplane came to rest against another jetway. Four of the 37 passengers sustained serious injuries. The airplane was destroyed by the fire and impact. The National Transportation Safety Board determines that the probabie cause of this accident was the improper installation of the high-pressure fuel filter cover that allowed a massive fuel leak and subsequent fire to occur in the right engine nacelle. The improper installation probably occurred at the engine manufacturer, buwever, the failure of atrlirie maintenance personnel to detect and correct the improper installation contributed to the accident. Alsc contriouting to the accident was the loss of the right engine center access panels from a fuel explosion that negated the fire suppression system and allowed hydraulic line burg through that in turn caused a total loss of airplane control on the ground.

[CONCLUSIONS > FINDINGS] The flightcrew and flight attendant were well rested before the flight, and there were no indications of chronic or stress-related factors that would have affected their performances. After takeoff, a loss of torque occurred on the right engine due to a drop in fuel pressure caused by a massive fuel leak from the high-pressure fuel filter cover. the flight was handled by air traffic control in accordance with applicable air traffic control procedures, and ATC response to the emergency was coramendable. The flight attendant’s instructions to passengers were concise and accurate, and her actions were commendable and instrumental in preventing more serious injuries. When the landing gear was lowered, a fire broke out in the right engine nacelle/right wheel well that subsequently rendered both the left and right hydraulic systems inoperative. The starter generator, located in the right engine compartment, had an improperly installed brush access cover. It could not be determined if this was the ignition source of the fire. The initial explosive force of the fire blew one of the engine cowl panels off and the other open and rendered the engine fire suppression system ineffective. Airplane control began to deteriorate in the air because of the loss of rudder control and roll spoilers on short final approach due to the burn through of hydraulic lines. Following touchdown, all airplane control was lost due to the loss of normal brakes, emergency brakes, nosewheel steering, and rudder control, During the emergency, the flightcrew performed commendably and exhibited coordinated crew interaction in accordance with good CRM concepts which mitigated the seriousness of the emergency. The rapid resporise of the aircraft rescue and firefighting perarnel was commendabie and instsumental in preventing fatalities 34 16. The shoulder harness and jumpseat hold-up strap in the cockpit of N819PH were worn beyond safe limits. fa : 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident / was the improper installation of the high-pressure fuel filter cover that allowed a massive fuet leak and subsequent fire to occur in the right engine nacelle. The improper installation probably occurred at the engine manufacturer; however, the failure of airline maintenance personnel tr detect and correct the impreper instatlation contributed to the accident. Also contributing to the accident was the loss of the right engine center access panels from a fuel explosion that negated the fire suppression system and allowed hydraulic line burn-through that in turn caused a total loss of airplane control on the ground. —

[ANALYSIS] Using this technique, no evidence ai ground spoiler activation was found on either wing in any ef the key video frames examined on the accident video tape 2. ANALYSIS 2.1 General The captain, first officer, and flight attendant aboard Horizon Air flight 2658 were trained and qualified for the flight in accordance with company policy and FAA regulations. The Safety Board also metes that Horizon Air has an established cockpit resource management (CRM) training program. The flightcrew’s actions during this accident illustrated familiarity with the concepts of this training, in addition, the flight altendant’s instructions lo passengers to take one of two brace positions (due to the seating arrangement of the airplane) were detivered correctly before touchdown. Further, her repeated insistence that the passengers remain in the braced position while the airplane rolled across the ramp and inte the jetways was important in preventing more serigUs Injuries. FAA ait traffic control personnel in the Seattle tower and approach control facilities performed their duties in a timely and appropriate manner during the accident sequence. During the first phase of the incident, after the flightcrew notified the tower that they were returning to land fwith no amplifying comments), the controller sequenced the airpiane into landing traffic according to established procedures Shortly thereafter, the local controiler ordered emergency personnel into position, even though he knew aniy that the airplane was returning for unknown reasons. Although the flightcrew had not declared an emergency at that point, the controler initiated an emergency equipment response solely as a safety precaution. Because the incident evolved from a simple precautionary landing into a catastrophic in-flight fire less than a minute before touchdown, the centraller’s actions in alerting the emergency crews resulled in a timely response and effective evacuation of the passengers and crew The effectiveness of the aircraft rescue and firefighting activities of the POSFD was alsa noteworthy. The fire that engulfed the right engine nacelle area was extinguished by 1839, within 7 minules after touchdown. in the video tape of the accident sequence, several firetrucks reversed their direction after the plane touched down, and in order ta be in good pasition to put out the fire and beqin passenger rescie as soon as possible, the firetrucks followed flight 2658 across the ramp when the crew fost control of the airplane. Tie rapid response of the emergency personne! wa, instrumental in saving the life of the passenger in seat iE who sustained a lacerated aorta the rescue of this passenger began before the fire was extinguished. 2.2 The Right Engine Fuel Leak and Fire The Safety Board determined that the cause of the fuel leak on the accident flight was the improperly installed fuel filter cover on the right engine high-pressure fuel pump. The Board believes that repeated high-pressure fuel pressurizations of the unsecured fuel filter cover allowed the neoprene o-ring to distort and extrude into a position so that it allowed high-pressure fuel to be channeled to a vent and drain hole on the filter housing and thereafter overboard into the nacelle. The distorted o-ring and its position in relation to the vent and drain hole appeared on radiographs before the filter cover was removed.

[ANALYSIS] Following the successful operational test of the left engine-driven hydraulic pump at the Vickers facility, this unit was disassembled. During the disassembly, a broken control spring guide was discovered. An analysis of this anomaly revealed that the broken guide would have tended to bias the pump toward maximum output, if the guide had interfered with the spring compression. Therefore, it was concluded that the engine-driven pump was functioning mormally until it cavitated while the airplane was on short final approach. The structure and orientation of the dimple overstress failures observed on the pump drive shaft end were corsistent with shearing oversiress forces. This type of failure is also consistent with the sudden stopping of the propeller geart‘ain during the impact sequence, rather than sudden stopping of overloading of the pump itself during flight. 2.4 Aircrew Actions The flightcrew noted nothing out of the ordinary during the preflight inspections of the exterior of the airplane. According to deHavilland and Horizon Air procedures, there is no leguirement jor aircrew inspection of the interior of the engine compartments during preflight activity, fhe entire incident involving flight 2658 spanned 6 minuies--when the initial partial power loss occurred at 1826:30 Lo impact witn jelway &tt at 1832:30. Until ihe fire broke out in the right enaine area, the flightcrew was confronted with am unexplained loss of right engine torque with no other associated problems. The Safety Board concludes that their actions in assessing the loss of power and its effect on the safe recovery of the airplane were appropriate and indicative of good CRM. The Safety Board notes that comments from the captain during the initial power floss such as: “Okay, hela me watch the airspeed there’; “Have that [emergency}} checklist standing by": and "Cikay, let's analyze {{for] anything else..." are good examples of a captain entisting the aid and knowledge of his first officer. Also, the captain's instruction to the firs. officer to advise the flight attendant that they were returning to the airport and the fact that he later double-checked that this was done insured that all three crewmembers were involved in the attempt to recover the NOPE px eRe EMME HDS THEE DENTE RMT Cee WrQreme ye Serine Merten een Uet We TiWtpRrne Ane ATW Zee neds. Pubiancn rane omer tat re is dans saad a= Ae rig A: aa Bi. BE SET TEN ee Tet Re aie ee Gl eee te Ee eee ee eG Toe On EE EE OES SE BIS og PEE : owes airplane safely. From the onset of the emergency, the captain treated the situation as a team effort.

[ANALYSIS] The pretouchdown loss of rudder coritrol, automatic ground spoiler activation, nose-wheel steering, pilot braking capability, and thrust from one engine precluded almost ail ability to steer the airplane. Conceivably, the airplane heading could have been changed by varying thrust on the operating left engine; however, such an action could have resulted in an increase in ground speed if pasitive thrust was applied, and the varying thrust possibly could have resulted i an inadvertent collision with other objects, such as taxiing or parked airplanes or the termingt building. During a postaccident interview, the captain stated that at the time, he considered cohision with the lightly constructed jetways a better option than a collision with the terminal building. 2.5 Airplane Design 2.5.1 Engine Fire Suppression versus Engine Cowl Design The Safety Board is very concerned that the effectiveness of the efigine fire suppression system was negated by apparent flaws in the design of the cowl and cov latches on the deHavilland DHC-8. uring this accident sequence, the left cowl on the right engine was blown off the nacelle when the fuel pooled in the nacelle ignited, Although it could n@t be determined positively, the right cowl on that engine probably was blawn open during the ‘nitial explosion and fell off the nacelle during impact with the jetways. When the {{irst officerfactivated the fire bottles an the engine shortly after the fire broke aut the fire suppressant wa expetied quickly onte and around aoe 2 re 31 an essentially uncowled engine to no apparent avail. With no cowls to contain the fire suppressant, the fire suppressant system was rendered ine‘fective. Following the accident, the center accass panels from the right engine were examined. Both panels were bowed out and except for one corner of the right panel, exhibited no fire damage or sooting. All latches on the panels were latched and undamaged. It was apparent that the outward force of the fuel explosion bowed and buckled the panels so that the latches could no longer hold the center access panels to the nace'le, The Safety Board is aware of another instance of apparent center access panel latch failure on another Horizon Air DHC-8. On June 19, 1987, aircraft N813PH experienced a right engine fire due to a leaking fuel line. However, in this instance, the center access panels remained attached hut in a loosened state, and the fire suppression system was effective. The Safety Board is pleased to note that deHavilland is exploring means to enhance the effectiveness of the engine cowls to preclude their toss during engine fires. Although an evaluation of the DHC-8 engine cow! design and installation revealed that they meet the requirements of the regulations, the Safety Board believes that the regulations should be reviewed to determine whether more stringent requirements are necessary.

[ANALYSIS] The exact I source of ignition could not be determined positively. The misplaced starter generator brush access cover on the right generator conceivably could have been a factor in the ignition because it may have allowed a combustible fuel/air mixture to accumulate in the area of the generator brushes. There is also another clear, unshielded path to the brush/armature area. Near the top of the starter, generator electrical leads progress into the generator armature and brush area. There isan open gap at this location which is about 1 foot closer to the fuel leak than the brush access cover. Therefore, in spite of the mispositioning of the access cover, there was another open path to an ignition source. ; Following the accident, on June 20, 1988, Lucas Aerospace Power Equipment Corporation issued &@ Service Information Letter 23088-00X-03 that outlined the correct installation of the starter-generator brush access covers on 23088 series generators. The Service Information Letter also recommended that any new or overhauled starter-generators be checked for correct brush cover installation before being piaced on engines. On July 22, 1988, Lucas Corporation issued Service Information Letter 23088-00X-04 that recommended a procedure for sealing the open gap associated with the electrical leads on 23088 series generators. This procedure was recommended to be accornplisiied at the earliest opportunity. On July 26, 1988, Transport Canada issued AD CF-88-15 that mandated compliance with these two Lucas Service information Letters. On September 2, 1988, FAA Af 88-18-12 became effective. This AD also called for mandatory compliance with the two Lucas Service Information Letters. Another possible ignition source could have been the engine exhaust pipe. Atomized, fuel could have been drawn into the cwaling air shroud surrounding the exhaust pipe. The area where this cooling air originated contained a large amount of accumulated fuel, 2.3 The Loss of Control on the Ground The Safety Board noted that 1m accordance with accepted airplane design practices, a fire and subsequent shutdown of one engine on a twin-engine airplane should not have caused the detertoration and suissequent loss of airplane control. The Board concluded that afl systerns that would have aided in stopping N&19PH on the ground after touchdown were disabled by the fire. 2.3.1 The No. 2 (Right) Hydraulic System Following the outbreak of the fire, the pilots immediately shut the right engine down in accordance with their emergency training. During a simple right er-yjine shutdown (with no other associated problems), the following components, which could only receive hydraulic pressure from the right engine-driven hydraulic pump or the Na 2 eiectrical-standby hydraulic pump, would be disabled: The inboard and outboa:d ground spatlers These wing-mounted automatically activated panels Normaily activate on touchdown and aid in airplane control by destroying lifton the wings and by acting as air brakes. The outboard roll spoilers.

[ANALYSIS] The flight attendants had completed ValuJet’s FAA-approved flight attendant training program. Visual meteorological conditions prevailed, and weather was not a factor in the accident. The evidence indicated that the accident airplane was equipped and maintained in accordance with Federal regulations and approved procedures. At the time of the accident, the three MEL [minimum equipment list] items and the one CDL [configuration deviations list] item had been deferred in accordance with approved lists; of these four items, the inoperative service interphone system will be discussed in section 2.4.3, and the remaining MEL/CDL items were not related to the accident sequence. There was no evidence of preexisting mechanical malfunctions or other discrepancies in the airplane structure, flight control systems, or powerplants that would have contributed to the accident. Based on the observed damage to rotating parts, both engines were developing power at ground impact. Evidence from the CVR revealed that about 6 minutes after takeoff from Miami, the crew of flight 592 became aware of a fire in the passenger cabin. Approximately 10 minutes after takeoff, flight 592 crashed into the Florida Everglades. The accident was not survivable. The catastrophic impact and destruction of the airplane precluded complete recovery of all airplane components. However, the wreckage that was recovered provided evidence of fire damage throughout the majority of the forward cargo compartment and areas of the airplane above it, with the most severe fire damage found in the ceiling area of the forward part of this compartment. Other areas of the airplane did not show significant fire damage, including the cockpit and the electronics compartment of the airplane located beneath the cockpit. The airplane’s electrical system was examined for indications as to what caused the electrical problems initially noted by the flightcrew. However, because so much of the wiring ran adjacent to the cargo compartment, and because so many of those wires were severely damaged, the source of those electrical anomalies could not be isolated. Examination of the heat-damaged wire bundles and cables revealed no physical evidence of short circuits or of burning that could have initiated the fire. Further, the heat and fire damage to the interior of the cargo compartment was more severe than the damage to the exterior, consistent with the fire having been initiated inside the cargo compartment. (The cargo compartment liner, which was designed to keep a fire contained within the cargo compartment, would also have functioned to keep an externally-initiated fire out of the compartment.) Finally, the heat-damaged wire bundles were not routed near the breached area of the cargo compartment, whereas the boxes containing the oxygen generators were loaded into the area directly beneath 101 the breached area of the cargo compartment. Thus, the electrical system was not a source of ignition of the fire. The investigation revealed that shortly before flight 592’s departure from Miami, five boxes of unexpended chemical oxygen generators and three tires (two of which included wheel assemblies)119 were loaded into the forward cargo compartment in the area where the fire damage was the most severe.

[ANALYSIS] Within 12 seconds, the captain reported an electrical problem, and at 1410:25, there were voices shouting “fire, fire, fire” in the passenger cabin. In the Safety Board’s fire tests,120 a main gear tire that had been inflated to 50 psi ruptured 16 minutes after the first oxygen generator was activated, when the fire destroyed 9 of the 12 tire sidewall plies. Because the tires in the accident airplane were loaded just forward of the cargo door, the tires would have been located just above the set of left static ports. The FDR altitude and speed data are based on readings from the left alternate static port, which is located on the left side of the fuselage at FS 341 between longerons 26 and 27, indicating that the unidentified sound on the CVR and the FDR anomaly at 1410:03 were most likely caused by the rupture of an inflated tire in the forward cargo compartment after the tire was partially burned through by the fire. Based on this sequence of events, the investigation analyzed when the fire on board the accident flight might have been initiated. Activation of a generator would have been most likely to occur during an event that could cause movement or jostling of the contents of the boxes. Accordingly, the Safety Board considered whether the fire might have been started as a result of one or more generators being activated during the loading process, which likely ended before 1340:29 when the passenger safety briefing was recorded on the CVR. The tire ruptured more than 30 minutes later. The Safety Board also considered whether the fire could have resulted from an oxygen generator being activated during the takeoff roll, which began about 1403:34. However, this was only 6 to 7 minutes before the tire ruptured. The Safety Board recognized that several factors could have affected the rate at which the accident fire developed, as compared to that seen in the Safety Board’s fire tests, including the relatively airtight design of the accident cargo compartment, the possible actuation of multiple oxygen generators, and the presence of more combustibles near the actuated generator than there were in the fire tests.121 Therefore, based on the general timing information obtained from the fire tests, the Safety Board concludes that one or more of the oxygen generators likely were actuated at some point after the loading process began, but possibly as late as during the airplane’s takeoff roll. Given that the fire had progressed sufficiently to cause a tire in the forward cargo compartment to rupture at 1410:03, the investigation examined why there was not an earlier indication of smoke and/or fire in the cabin than the first audible report at 1410:25. Several factors might account for the lack of warning from smoke earlier in the fire sequence.

[ANALYSIS] Thus, the electrical system was not a source of ignition of the fire. The investigation revealed that shortly before flight 592’s departure from Miami, five boxes of unexpended chemical oxygen generators and three tires (two of which included wheel assemblies)119 were loaded into the forward cargo compartment in the area where the fire damage was the most severe. The investigation further found that safety caps were not installed over the percussion caps that start a chemical reaction in the oxygen generators; lanyards for the retaining pins for the percussion caps’ spring-loaded actuation mechanism were not secured on several generators; and the generators were not packaged adequately to prevent generators from striking the actuation mechanism or dislodging retaining pins on adjacent generators. Based on the results of the Safety Board’s fire tests on chemical oxygen generators that were conducted near Atlantic City, New Jersey, after the accident, the physical evidence of fire damage in the forward cargo compartment of the accident airplane, and the lack of other cargo capable of initiating a fire in the forward cargo hold, the Safety Board concludes that the activation of one or more chemical oxygen generators in the forward cargo compartment of the airplane initiated the fire on ValuJet flight 592. The Safety Board’s analysis, therefore, first examines the accident sequence, including the initiation and propagation of the onboard fire, and the adequacy of air carrier and FAA efforts to minimize the hazards posed by fires in cargo compartments of commercial airplanes. The analysis also explores the pilots’ performance and actions when they became aware of the fire shortly after takeoff from Miami, and the adequacy of smoke protection equipment and smoke evacuation procedures aboard air carrier aircraft. The analysis then examines the circumstances surrounding the shipment of the oxygen generators and the procedures for shipping company material and hazardous materials. The analysis also evaluates concerns raised regarding the adequacy of the FAA’s hazardous materials program; ValuJet’s outsourcing of maintenance and training activity; the company’s oversight of its contract maintenance facilities; and the FAA’s oversight of ValuJet and ValuJet’s contract maintenance facilities. Finally, the analysis addresses the adequacy of ValuJet’s procedures for manifesting lap children. 2.2 Propagation and Detection of Fire The first indication of a problem during the accident flight occurred at 1410:03, approximately 6 minutes after flight 592 took off from Miami, when the CVR recorded an unidentified sound, which prompted the captain to ask “What was that?” Simultaneously, an anomaly in the FDR altitude and airspeed parameters occurred consistent with a static pressure 119 Although the stock clerk and others who handled the tires believed that all three were mounted on wheel assemblies, evidence recovered at the accident site (including several pieces of oxygen generators and other debris found inside the unruptured main tires) indicated that one of the main gear tires was not mounted on a wheel assembly. 102 increase of about 69 psf. Within 12 seconds, the captain reported an electrical problem, and at 1410:25, there were voices shouting “fire, fire, fire” in the passenger cabin.

[ANALYSIS] The task force is working on the manifest issues, including the subject of lap children. A report from the Secretary of Transportation is due to Congress by October 8, 1997. 133 3. CONCLUSIONS 3.1 Findings 1. The flightcrew was properly certificated and had received the appropriate training and offduty time prescribed by the Federal regulations. 2. There was no evidence that any preexisting medical condition affected the flightcrew’s performance. 3. The flight attendants had completed ValuJet’s Federal Aviation Administration-approved flight attendant training program. 4. Weather was not a factor in the accident. 5. The accident airplane was equipped and maintained in accordance with Federal regulations and approved procedures, and there was no evidence of preexisting mechanical malfunctions or other discrepancies in the airplane structure, flight control systems, or powerplants that would have contributed to the accident. 6. The activation of one or more chemical oxygen generators in the forward cargo compartment of the airplane initiated the fire on ValuJet flight 592. One or more of the oxygen generators likely were actuated at some point after the loading process began, but possibly as late as during the airplane’s takeoff roll. 7. Even if the fire did not start until the airplane took off, a smoke/fire warning device would have more quickly alerted the pilots to the fire and would have allowed them more time to land the airplane. 8. If the plane had been equipped with a fire suppression system, it might have suppressed the spread of the fire (although the intensity of the fire might have been so great that a suppression system might not have been sufficient to fully extinguish the fire) and it would have delayed the spread of the fire, and in conjunction with an early warning, it would likely have provided time to land the airplane safely. 9. Had the Federal Aviation Administration required fire/smoke detection and fire extinguishment systems in class D cargo compartments, as the Safety Board recommended in 1988, ValuJet flight 592 would likely not have crashed. 10. Given the information available, the ramp agents’ and flightcrew’s acceptance of the company materials shipment was not unreasonable. 11. ValuJet’s failure to secure the cargo was not unreasonable. 134 12. The loss of control was most likely the result of flight control failure from the extreme heat and structural collapse; however, the Safety Board cannot rule out the possibility that the flightcrew was incapacitated by smoke or heat in the cockpit during the last 7 seconds of the flight. 13. Only a small amount of smoke entered the cockpit before the last recorded flightcrew verbalization at 1411:38, including the period when the cockpit door was open. 14.

[ANALYSIS] After takeoff, the captain was operating the controls of the airplane and the first officer was handling communications with air traffic controllers. Based on the CVR recording, the flightcrew’s performance was appropriate during the portions of the flight that preceded the crew’s first awareness that a problem existed. 2.4.2 Flightcrew Decisions and Actions During the Emergency Beginning at 1410:12, the flightcrew noted and verbalized concerns about electrical problems. At 1410:22, the captain stated, “We need, we need to go back to Miami.” This was followed 3 seconds later by shouts in the background of “fire, fire, fire.” Seven seconds later, as the first officer transmitted a request to ATC for clearance to return to Miami (and before receiving clearance from ATC), the airplane leveled off and began to descend. 122 49 CFR 175.81(a) states, “[p]ackages containing hazardous materials must be secured in an aircraft in a manner that will prevent any movement in flight which would result in damage to or change in the orientation of the packages.” 106 Based on the shouts from the passenger cabin recorded on the CVR cockpit area microphone at 1410:25 and the comment 2 seconds later, “we’re on fire, we’re on fire,” it should have been clear to both flightcrew members that a very serious emergency situation existed in the cabin. Although the captain decided immediately to return to Miami and initiated a descent, for the next 80 seconds the airplane continued on a northwesterly heading (away from the Miami airport) while the flightcrew accepted ATC vectors for a wide circle to the left and a gradual descent back toward Miami. The Safety Board evaluated the electrical system, engine, and flight control malfunctions that occurred in the 80 seconds during which the airplane continued northwestward away from MIA. The electrical problems that first made the flightcrew aware of the emergency (at 1410:12) likely were the result of insulation burning on wires in the area of the cargo compartment. Electrical system wiring is routed outside of the cargo compartment of the DC-9, in accordance with 14 CFR Part 25.869, which requires the wiring not to be located against the cargo compartment liner and to incorporate a high temperature insulation. Therefore, the flightcrew’s comments about the electrical problems indicate that the fire had probably already escaped the cargo compartment by 1410:12. (However, it probably had not yet burned through the cabin floorboards.) The flightcrew comments recorded by the CVR from 1410:12 through 1410:22 reflect the pilots’ concerns about and attention to these electrical problems. It is possible that these concerns continued to occupy some of the pilots’ attention during the initial period of their attempt to return to the ground. Another malfunction began at 1410:26, just as the shouts from the cabin would have alerted the flightcrew to the seriousness of the fire there. According to FDR data, while the left engine remained at its previous EPR setting, the right engine’s EPR decreased to the flight idle value.

[ANALYSIS] Given that the fire had progressed sufficiently to cause a tire in the forward cargo compartment to rupture at 1410:03, the investigation examined why there was not an earlier indication of smoke and/or fire in the cabin than the first audible report at 1410:25. Several factors might account for the lack of warning from smoke earlier in the fire sequence. First, the 120 The tests were not designed to be an exact replication or simulation of the circumstances of this accident, but were conducted to learn about the overall nature of a fire initiated by an oxygen generator and fed with high concentrations of oxygen released from additional oxygen generators. Because the investigation could not conclusively determine the exact physical arrangement of the generators in each box, the exact size of the boxes, the exact arrangement of the boxes and tires in the cargo compartment, or how many oxygen generators were initially activated, and because of differences between the test chamber and the accident cargo compartment, the Safety Board recognizes that the test results might differ somewhat from what occurred on the accident airplane. 121 See also the factors outlined in footnote 120, above. 103 cargo compartment liner is designed to limit the amount of ventilation to and from the cargo compartment; consequently, so long as the liner is intact, the smoke will not readily escape into the passenger compartment. Second, any smoke that did escape would not have readily entered the air flow in the passenger cabin, which comes from overhead and down into the area between the airplane outer skin and the cargo liner, then moves aft and exits through the outflow valve. Third, the oxygen generators would have initially fed the fire with an abundance of oxygen, tending to minimize the amount of smoke and resulting in a very rapidly developing fire. All of these factors in combination most likely prevented any noticeable migration of smoke from the forward cargo compartment into the passenger cabin or cockpit until relatively late in the development of the fire. Although black soot deposits on some of the overhead luggage compartments indicate that black smoke ultimately reached the passenger cabin, this smoke probably did not reach the passenger cabin until after the fire had breached the cargo compartment ceiling. Because the cargo compartment where the fire occurred was a class D cargo compartment and was not equipped (nor was it required to be equipped) with a smoke detection system, the cockpit crew of ValuJet flight 592 had no way of detecting the threat to the safety of the airplane from the in-flight fire until smoke and fumes reached the passenger cabin. Further, because the cargo compartment was not equipped (nor was it required to be equipped) with a fire suppression system, the cockpit crew had no means available to extinguish or even suppress the fire in the cargo compartment. If the fire started before takeoff, and a smoke/fire detection warning device had activated, the flightcrew most likely would not have taken off. However, the Safety Board concludes that even if the fire did not start until the airplane took off, a smoke/fire warning device would have more quickly alerted the pilots to the fire and would have allowed them more time to land the airplane.

[ANALYSIS] If the fire started before takeoff, and a smoke/fire detection warning device had activated, the flightcrew most likely would not have taken off. However, the Safety Board concludes that even if the fire did not start until the airplane took off, a smoke/fire warning device would have more quickly alerted the pilots to the fire and would have allowed them more time to land the airplane. Further, the Safety Board concludes that if the plane had been equipped with a fire suppression system, it might have suppressed the spread of the fire (although the intensity of the fire might have been so great that a suppression system might not have been sufficient to fully extinguish the fire) and it would have delayed the spread of the fire, and in conjunction with an early warning, it would likely have provided time to land the airplane safely. Although class D cargo compartments are designed to suppress fire through oxygen starvation, this accident and events before this accident illustrate that some cargo, specifically oxidizers, can generate sufficient oxygen to support combustion in the reduced ventilation environment of a class D cargo compartment. The in-flight fire on American Airlines flight 132, a DC-9-83, on February 3, 1988 (see section 1.6.3.2), clearly illustrated the need for systems that would provide flightcrews with the means to detect and suppress fires in the cargo compartments of airplanes. As a result of its investigation of that accident, the Safety Board recommended that the FAA require fire/smoke detection and fire extinguishment systems for all class D cargo compartments (Safety Recommendations A-88-122 and -123). As recently as August 1993, although the FAA had investigated several incidents of fires that were initiated as a result of oxidizers in the cargo compartments of airplanes, the FAA responded to Safety Recommendations A-88-122 and -123 stating that fire/smoke detection and fire extinguishment systems were not cost beneficial, that it did not believe that these systems would provide a 104 significant degree of protection to occupants of airplanes, and that it had terminated its rulemaking action to require such systems. The Safety Board concludes that had the FAA required fire/smoke detection and fire extinguishment systems in class D cargo compartments, as the Safety Board recommended in 1988, ValuJet flight 592 would likely not have crashed. Therefore, the failure of the FAA to require such systems was causal to this accident. The crash of ValuJet flight 592 prompted the FAA to state in November 1996 that it would issue an NPRM by the end of the summer of 1997 to require, on about 2,800 older aircraft, the modification of all class D cargo compartments to class C compartments, which are required to have both smoke detection and fire extinguishment systems. The accident also prompted the ATA to announce in December 1996 that its members would voluntarily retrofit existing class D cargo compartments with smoke detectors. As of the date of this report, the Safety Board is unaware that any airplanes have been modified and are in service.

[CONCLUSIONS > FINDINGS] Because of the lack of information regarding products approved for transportation by the Bureau of Explosives, Research and Special Programs Administration cannot adequately ensure that these products are being packaged and shipped safely in the transportation environment. 46. ValuJet did not follow its internal procedures for boarding and accounting for lap children. 47. It is essential that air carriers maintain easily accessible and accurate records of the names of both ticketed and unticketed passengers aboard their flights for retrieval in the event of an accident or other emergency. 3.2 Probable Cause The National Transportation Safety Board determines that the probable causes of the accident, which resulted from a fire in the airplane’s class D cargo compartment that was initiated by the actuation of one or more oxygen generators being improperly carried as cargo, were (1) the failure of SabreTech to properly prepare, package, and identify unexpended chemical oxygen generators before presenting them to ValuJet for carriage; (2) the failure of ValuJet to properly oversee its contract maintenance program to ensure compliance with maintenance, maintenance training, and hazardous materials requirements and practices; and (3) the failure of the Federal Aviation Administration (FAA) to require smoke detection and fire suppression systems in class D cargo compartments. Contributing to the accident was the failure of the FAA to adequately monitor ValuJet’s heavy maintenance programs and responsibilities, including ValuJet’s oversight of its contractors, and SabreTech’s repair station certificate; the failure of the FAA to adequately respond to prior chemical oxygen generator fires with programs to address the potential hazards; and ValuJet’s failure to ensure that both ValuJet and contract maintenance facility employees were aware of the carrier’s “no-carry” hazardous materials policy and had received appropriate hazardous materials training. 138

[PROBABLE CAUSE] AE. the pilot stopped the airplane, the airport fire department began firefighting operations. Flight attendants and passengers opened the left and rig:ht forward doors, the left forward overwing exit, and the right forward and aft overwing exits. About 60 to 90 seconds after the exits were opened, a flash fire engulfed the airplane interior. While 18 passengers and 3 flight attendants exited through the forward doors and slides and the three open overwing exits to evacuate the airplane, the captain and first officer exited through their respective cockpit sliding windows. However, 23 passengers were not able to get out of the plane and died in the fire. The airplane was destroyed. As a result of its investigation, the Safety Board determined that •.. the probable causes of the accident were a fire of undetermined origin, an underestimate of fire severity, and conflicting fire progress information provided to the captain. Contributing to the severity of the accident was the flightcrew's delayed decision to institute an emergency descent. APPENDIX H -2In its investigation, the Safety Board found sufficient evidence to substantiate the conclusion that 8. fire propagated through the amenities section of the aft lavatory and had burned undetected for almost 15 minutes before the smoke was first noticed. Although, the Board was not able to identify the origin of the fire, it was able to eliminate several possibilities and identify its general location. The evidence disclosed that the fire propagated through the lower part of the amenities section of the lavatory vanity. Since the direction of airflow below the vanity and above the toilet waste tank permitted venting of smoke, fumes, and hot gases away from the cabin and open area of the lavatory, the fire went undetected for almost 15 minutes until it began to penetrate the lavatory liner and rise behind and outboard the liner, subsequently penetrating the sidewall and ceiling seams of the lavatory liner. The Petitioner acknowledged that it has no new evidence to present with regard to the probable cause of the accident. However, the Petitioner presents three principal assertions for the Safety Board's review: first, the Safety Board did not properly assess the available evidence regarding the time factor in the captain's decision to descend and land the airplane; second, the use of the word "misconstrue" in finding No. 8 carries a connotation of error on the part of the captain; and third, the contributing statement of the probable cause suggests that the captain's decision to institute an emergency descent was delayed longer than a prudent captain would have waited under similar circumstances. With regard to the first issue, the Petitioner referenced page 61 of the Board's report, which stated, in part, that "The Safety Board believes that a precautionary emergency descent shpuld have been initiated as soon as it became evident that the fire had not been visually located and could not be attacked directly with extinguishant.

[ANALYSIS] FAA regulations and company policies and procedures. The flightcrew was qualified and certificated properly and the flight attendants were qualified for the flight. Each flight and cabin crewmember had received the training and off-duty time prescribed by Canadian regulations. There was no evidence of any preexisting medical or psychological conditions that might have affected thfi! performance of the flight and cabin crews. Involved air traffic controllers were certificated properly, and each controller had received the training and off-duty time prescribed by FAA regulations. Accordingly, the Safety Board directed its investigation to the ignition and propagation of the fire; to ATC procedures; to the performance of the pilots and flight attendants after the fire was discovered; and to factors which affected the survivability of the passengers and crewmembers. 2.2 Fire Ignition.-The evidence substantiates a conclusion that when the smoke was detected by the flight attendants, there was a fire located within the vanity and/or the toilet shroud in the aft lavatory. Therefore, the Safety Board tried to identify all possible ignition sources in this area. Given the location of the fire at the time the smoke was discovered, the Safety Board identified five possible ignition sources: an incendiary or explosive device; deliberate ignition; a burning cigarette; the toilet flush motor; or the flush motor electrical harness. In addition to these five, the arcing damage found on the feeder cables of the left and right a.c. generators and the maintenance history of the airplane a.c. generating system led the Safety Board to investigate a sixth possible source of ignition-the generator feeder cables which were routed beneath the floor of the aft lavatory. Based on the examination of the physical evidence and the results of the FBI laboratory analysis, the Safety Board concluded that neither an explosive nor incendiary device was involved. Also, there was no evidence that the fire was deliberately set. . Since the tests of the materials used in the aft lavatory showed that they met the fire resistance criteria contained in 14 CFR 25, it would have been virtually impossible for either a lighted cigarette or sparks produced by electrical arcing to ignite the materials used in the construction of the lavatory. In order to ignite the lavatory partitions and walls, some combustible material capable of sustaining high temperature combustion for the amount time necessary to ignite the lavatory walls had to have burned. Therefore, in investigating the possibility that the fire was ignited by a burning cigarette, the Safety Board focused on two areas below the vanity which could have contained combustible materials and into which a cigarette might have fallen-the sink area containing the trash chute and receptacle and the adjoining- amenities section. Since the lavatory trash receptacle was the most logical place for combustible material to collect and since a burning cigarette could fall down the trash chute into the receptacle, the damage in this area was evaluated. Had a cigarette started a fire in the receptacle, the only propagation path out of the receptacle would have been the trash chute.

[CONCLUSIONS > FINDINGS] Crewmember reports that the fire was abating misled the captain about the fire severity and he delayed his decision to declare an emergency and descend. 9. Because of the delayed decision to descend, the airplane lost the opportunity to be landed at Louisville. Had the airplane been landed at Louisville, it could have been landed 3 to 5 minutes earlier than it actually did land at Cincinnati. The delayed decision to descend and land contributed to the severity of the accident. 10. A faulty ATC handoff did not delay significantly Flight 797's landing at Greater Cincinnati Airport. 11. The fire consumed the lavatory walls, propagated into the ceiling, and then began to move forward. Smoke, toxic fumes, and heated gases began to enter the cabin, spread forward, and collect along the ceiling of the cabin. 12. The flight attendants' passing out wet towels to the passengers and instructing them to breathe through the towels or through articles of clothing aided in the survival of some of the passengers. 13. The first officer turned off the air conditioning and pressurization packs in the belief that the airflow was feeding the fire. The resulting loss of circulation accelerated the accumulation of smoke, heat, and toxic gases in the cabin and likely decreased the time available for evacuation. 14. Three of the four overwing exit windows were opened by designated passengers who had been selected and briefed to open them by the flight attendants. 15. When the airplane stopped, smoke had filled the cabin and visibility within the cabin was almost nonexistent 2 to 3 feet above the cabin floor. -7116. A flashfire occurred within the cabin within 60 to 90 seconds after the doors and overwing window exits were opened. Flames from this fire were not evident until after the survivors had left the airplane. Flames from the original fire never were evident within the airplane or to persons on the ground. 17. This was a survivable accident. 3.2 Probable cause The National Transportation Safety Board determines that the probable causes of the accident were a fire of undetermined origin, an underestimate of fire severity, and misleading fire progress information provided to the captain. The time taken to evaluate the nature of the fire and to decide to initiate an emergency descent contributed to the severity of the accident. 4. RECOMMENDATIONS On July 11, 1973, the Safety Board participated in the investigation of the Varig Airlines Boeing 707 accident near Paris, France, in which 124 persons died after a fire erupted in the rear lavatory.

[PROBABLE CAUSE] But, that is not the first reason why I went back tq the cockpit; the first thing I say after ....• I can't go back itls too heavy, I think \'ie'd better go down" Exhibit 12A, page 10. But we will come back to that part later. The report says in the same paragraph: "Since the flight attendant in charge and the first officer were not able to determine the location of the fire they were not able to assess the severity of the fire. Consequently••• , they provided the captain with an inadequate assessment of the fire's severity.··. That is easier said than done, and I quote from the following reports. Passenger Arnold Friedman (Exhibit 6A): "The male FIA had sprayed it with an extinguisher. Heavy brown smoke billowed forward. The first officer went aft several times." Passenger Harry Mosely: ..••. the noise of the CO2 fire extinguisher when triggered. This sound came from the back of the aircraft. He looked back and sa\'I the smoke." Flight attt=!ndant J. Davidson: "She ingested some smoke which made her feel dizzy... Mr. Benetti was standing back while I was trying to walk through the smoke and said in his report: "The smoke had an electrical acrid smell and burned his throath.·' and Mr. Benetti managed only to discharge part of the C02. So the flight attendant and I were not able to determine the location of the fire because we COULD NOT, NOT PHYSICALLY POSSIBLE, TOO MUCH SMOKE. This is where the assessment of our story stops short of reality. The report continues in paragraph 4: "The Safety Board bel ieves that a precautionary emergency descent should have been initiated as soon as it became evident that the fire had not been visually located and could be attacked directly with extinguishant. This became known at 19:04:04 when the first officer came forward from his first inspection of the aft cabin area, about 3 minutes before the decision to begin an emergency descent•..• Again, Exhibit 12A, page 10, I say "1 canlt go back now, itls too heavy (smoke). I think we'd better go down:'. So that is exactly what the report says we should have done and for the same reasons except for the "precautionary emergency descent~. I had in mind a descent, not a fuel saving descent of course because of the nature itself of the reason why •• ./3 APPENDIX G we were going down, but not an emergency descent also as I testified because I did not know enough.

[PROBABLE CAUSE] Again, all of this is very compatible with the lines of the report page 57, paragraph 4: .../5 APPENDIX G ··While an actual infl ight fire is an extremely rare occurence, all reports of smoke in the cabin must be regarded as potentially serious. However, such reports often turn out to be smoke from an overheated flushing motor or waste ignited by a discarded cigarette in a trash receptacle designed to contain a fire, conditions which are normally identified and corrected by flight attendants without further consequences. Therefore, the Safety Board realized that there is a need to evaluate the situation before deciding on the emergency action required.··. So I did not just stand there but went back with my smoke goggles to evaluate the last events, touched the door and knew immediately that we had a major problem and there was nothing we could do except to get on the ground and fast. I told the flight attendants to prepare for emergency landing and at 19:07:11 told the captain (Exhibit 12A): • "1 don't like ,...hat is happening, 1 think ,...eld better go down, O.K:·. The captain concurred and so it was between 19:04:07 (first report to the captain) and 19:07:11 (second report) as the development of the situation caused the decision to descend to be delayed, again, not knowing what would happen 'after. So the N.T.S.B. report (page 57, paragraph 4) goes on to say: "As a result, about 5 minutes 30 seconds elapsed between the time that no. 2 flight attendant told the captain there was a fire in the aft lavatory and his decision to begin emergency descent ..• however in t~i~ case the time to make the decision appears excessive given the circumstances.... The decision to initiate emergency descent was in fact made at 19:07:11 (Exhibit 12A), which is not 5 minutes 30 seconds but 4 minutes 31 seconds. At 19:08:12, the first Mayday was heard by ATC, the one minute difference was the time that I took to take my seat, declare Mayday twice, select 7700 on transponder, select emergency powel~ "on·' and finally the third f4ayday was heard (Exhibit 2A Ouimet statement). The first two Maydays were not heard because of electrical failures. The left electrical side failed before I arrived, and as 1 was calling the first Mayday, the right side failed at 19:07:41 (Exhibit lOA, page 3/ Exhibit 12A, page 14 OFDR and CVR fai1ed/ N.T.S.B. report page 4 paragraph 2). Therefore there was no power to the "Emet"gency Bus" where no. 1 radio is supplied and consequently I was unable to transmit.

[PROBABLE CAUSE] However, in this case, the time to make the decision appears excessive under the circumstances. The Safety Board believes that a precautionary emergency descent should have been initiated as soon as it became evident that the fire had not been visually located and could be attacked directly with extinguishant. This became known at 1904:07 when the first officer came forward from his first inspection of the aft cabin area, about 3 minutes before the decision to begin an emergency descent." APPENDIX G Mr. James Burnett -1002 December 20, 1984 We agree wholeheartedly that the emergency descent should have been initiated as soon as it became evident to the captain that the fire had not been visually located ~nd could not be combated directly. However, the evidence presented in the report indicates that this did not occur at 1904:01 as stated in the Board Report. On page 61 of the report, the Board states, "such reports, (of smoke in the cabin), often turn out to be smoke from an overheated flushing motor or waste ignited by a discarded cigarette in a trash receptacle designed to contain a fire, conditions which are normally identified and corrected by flight attendants without further consequences." At 1906:52, the captain, by his own testimony, and as indicated on page 3 of the Board report, still believed the fire to be in the lavatory trash bin, and he did not decide to descend at this time because, "I expected it (the fire) to be put out." He had been led to this decision by a series of positive reports with respect to the firefighting efforts of the crew. At 1904:16, only 9 seconds after the first officer had suggested that they go down, the flight attendant advised the captain, "you don't have to worry I think it's gonna be easing up." Seven seconds later, at 1904:23, the first officer advised, "Okay, it's starting to clear now." At 1905:35, another flight attendant advised, "Captain, your first officer wanted me to tell you that Sergio has put a big discharge of CO in the washroom, it seems ~o be SUbsiding all right." Some time after 1909:01, the flight attendant who gave the first report advised the captain, "Getting much better, okay." At 1906:52, the flight attendant advised the captain, "C02, it was almost Jfalf a bottle and it now almost cleared." At 1906:54, the captain responded, "Okay, thank you." Between, 1904:01 and 1906:54, the captain received 5 separate communications from three different crewmembers indicating that the fire was being brought under control and that the smoke was subsiding.

[PROBABLE CAUSE] While the extent to which this time interval contributed to the severity of the accident could be debated, the Safety Board believes that the criticality of an early decision under such circumstances is not in issue. The facts known to the captain at 1904:07 should have been sufficient indicators of a potentially serious hazard that a precautionary descent should have been made immediately. The first officer for his part should have realized that the acrid smelling smoke probably was the result of an electrical fire, associated it with the tripping of the aft lavatory flush motor circuit breakers 11 minutes earlier, and informed the captain accordingly. Despite the fact the flight attendant in charge told the captain at 1906:42 that, "••• even though I could not see the source but it's definitely inside the lavatory," he, in fact, observed thick curls of black smoke emerging through the seams of the aft wall of the lavatory and believed that the fire was not in the trash container -and had also informed the first officer of his belief. His observations and assessment of the fire danger were inconsistent with his report to the captain. It is the Board;s belief that the: flight attendant in charge should have realized that his fire fighting effort was ineffective and that the temporary subsiding of the smoke was misleading. Therefore, the Board concludes that these crewmember reports influenced the captain's decision to delay the initiation of a descent. The delay increased the time for the fire to propagate and the time that passengers were exposed to the toxic environment before the airplane could be evacuated. With regard to the second issue, the Petitioner disagrees with finding No.8 on page 70 that the captain misconstrued the fire progress reports from his crew. The Petitioner believes that the captain relied on his crew's reports, and states that IlThere is a vast difference between believing an erroneous communication, and misinterpreting a communication." Further, the Petitioner contends that the evidence does not support finding No.8. Therefore, the Petitioner requests that the last sentence of finding No.8 be changed to read, "Due to persistent reports from various crewmembers that the fire was abating, he delayed his decision to declare an emergency and descend." After reviewing finding No.8, the Safety Board agrees that the word "misconstrued" may have conveyed the wrong intent. The Board agrees that the captain was misled by the reports from crewmembers that the smoke was SUbsiding. However, the Board continues to believe that, in the interest of safety, the captain should have been more assertive in seeking the nature of the fire instead of accepting the nonspecific comments of the crewmembers as assurance that the situation was under control. Had the crewmembers positively stated that the fire source had not been identified, the captain might have taken action sooner to descend in preparation for an emergency landing. With regard to the third issue, the Petitioner does not agree that contributing to the severity of the accident was the flightcrew's delayed decision to institute an emergency descent.

[PROBABLE CAUSE] With regard to the first issue, the Petitioner referenced page 61 of the Board's report, which stated, in part, that "The Safety Board believes that a precautionary emergency descent shpuld have been initiated as soon as it became evident that the fire had not been visually located and could not be attacked directly with extinguishant. This became known at 1904:07 when the first officer came forward from his first inspection of the aft cabin area, about 3 minutes before the decision to begin an emergency descent." The Petitioner does not disagree with the Board's belief, but contends that the time at which this conclusion became apparent to the captain did not occur at 1904:07, and the Petitioner offers the following points from the Board's report to support its position: The Board acknowledged the occurrence of previous in-flight incidents involving lavatory fires which normally have been identified and corrected by flight attendants without further consequences; Trash receptacles in lavatories are designed to contain a fire; At 1906:52, the captain indicated that he believed the fire to be in the lavatory trash bin and that he did not decide to descend at this time because he expected the fire to be extinguished; The captain was misled by a series of reports from his crew indicating that the fire was under control and being extinguished. (These were five reports between 1904:07 and 1906:54 from three different crewmembers.) The captain sent the first officer aft to combat the fire 39 seconds after the time the Board believes it was evident to the captain that the fire could not be attacked directly with extinguishant; The captain would not have taken this action had he believed the fire was not accessible at that time. -3APPENDIX H The Petitioner therefore concludes that the captain did not become aware of the uncontrollable nature of the fire until 1907:11, when the first officer returned from the cabin and advised, III don't like what's happening, I think we better go down. Okay?" The descent was begun immediately after the first officer resumed his seat. All parties to the investigation were in agreement that the time factor involved contributed to the severity of the accident. The Board's rationale with regard to the delayed decision was that had the captain decided to descend immediately after being told by the first officer, at 1904:07, that, "... I can't go back now, it's too heavy, I think we'd better go down,1l the flightcrew could have conceivably landed the airplane 3 to 4 minutes earlier. While the extent to which this time interval contributed to the severity of the accident could be debated, the Safety Board believes that the criticality of an early decision under such circumstances is not in issue. The facts known to the captain at 1904:07 should have been sufficient indicators of a potentially serious hazard that a precautionary descent should have been made immediately.

[PROBABLE CAUSE] Female flight attendant at 19:05:36: ··..• it seems to be subsiding now". Flight attendant in charge at 19:06:52: ..... it is almost cleared". That change of situation is confirmed by many passengers' statements in Exhibit 6A. Arnold Friedman: "There was a short interval in which the smoke subsided". Glen W. Davis: ··•••and the smoke cleared for a minute or two". In Exhibit 6A, page 39, the Human Factors Group Chairman's factual report: "Severa1 passengers reported a short interval, estimated by one passenger to be 5 minutes, during which the smoke subsided. ". I believe that is enough evidence to confirm that at the time "Ie \'/ere deciding for a descent, there was a significant change in our situation: a decrease and complete disappearance of smoke in the cabin. The report fails to recognize that fact in spite of all the reports in the exhibits. What is called "conflicting fire progress", "underestimate of fire severity", "miscontrued reports" ~/ere just a simple observation, a noting of the facts by the crew. Of course, we did not have the benefit of knowing the future and we believed what we saw, and so it was for everyone on board. I remember when I went back for the second time that some people had resumed their drinking and talking. There are a lot of questions that go through your mind in a moment like this: Was the C02 effective? Maybe, the motor pump blew, produced smoke and died? Remember the pump was very present in my mind. Those were reasonable assumptions at the time, much more than the extraordinary set of circumstances that the report says happened (page 54, paragraph 5): "The momentary smoke abatement noted by the first officer' and fl ight attendant in charge between 19:04:16 and 19:04:23 (right at the time of our decision to descend) was probably attributable to the dilution effect of opening the lavatory door and discharging the C02 into the area, reclosing the lavatory door, and the almost simultaneous failure of the lavatory vent line and the flush and fill pipe connections and check valve, all of which increased the ventilation rate beneath the toilet shroud and accelerated the flow of smoke and gases to the area below the lavatory floor and overboard..·• It is the timing and this amazing sequence of events that led us to believe that the situation had improved, and so it was for everybody on board, and no one that was not there can pretend otherwise. Again, all of this is very compatible with the lines of the report page 57, paragraph 4: .../5 APPENDIX G ··While an actual infl ight fire is an extremely rare occurence, all reports of smoke in the cabin must be regarded as potentially serious.

[PROBABLE CAUSE] In an emergency descent, I am required in my seat and f thought that I could do more, maybe, with my goggles and the fact that I vias caught by surprise by the acridity of the smoke the first time and I would not be this tiole. I wanted to give it another shot so that I could see or even better fight the fire. Again from the cockpit voice recorder, Exhibit l2A, at 19:04:46 the captain says: ··Okay go back \'/henever you can but don I t get yourse1 f incapacitated.'· and my answer is: "no problem, no problem··. Not knowing what we know now, was it that unrealistic to think that way? During all that time, I do have the motor pump in my head as a possible cause ••• Exhibit 12A, transcript of the cockpit voice recorder, at 19:02:50 just after the flight attendant's report of a fire and before I left the cockpit, I say: "Got all the breakers pulled:·. Again, just before I left, I say: "You got all the breakers pulled out?" and the captain answers: ··The breakers are all pulled, yeah:·. In the cabin, the smell seemed to be coming very possibly from something electrical according to passengers and flight attendants' reports in Exhibit 6A. So I associated the fire with the lavatory pump as I testified in Cincinnati, but I could not convince myself since the breakers were all out and therefore the pump was supposed to be dead. So this is what happened between 19:02:40 and 19:04:07 and I am convinced that Captain Cameron would have gone down as I suggested because we worked all month (May) together and we had a very good professional relationship. " BUT, and the next 3 minutes raised even more harsh critici~m from the N.T.S.B. report, we did not commence descent and that is because at 19:04:16 Mr. Benetti says (report page 3, paragraph 3): •••••didn·t have to worry, I think it's gonna be easing up:' and at 19:04:23, I say to the captain: ··It is starting to clear now.··. • •• /4 APPENDIX G I never said not to descend, but I realize it implied just that, and the captain looked back and saw the cabin clear. Female flight attendant at 19:05:36: ··..• it seems to be subsiding now". Flight attendant in charge at 19:06:52: ..... it is almost cleared". That change of situation is confirmed by many passengers' statements in Exhibit 6A.

[PROBABLE CAUSE] It is very unfortunate that the news media did not quote the Board's comments that the captain exhibited outstanding airmanship without which the airplane and everyone on board would certainly have perished, but blared in headlines from coast to coast and in at least two countries, "Crew's delayed deciision causes accident." The crew's reputation was therefore destroyed by conclusions reached by the Safety Board which were not in accordance with the evidence in the Board's possession. We therefore respectfully request that the Safety Board change the last paragraph of the probable cause to indicate that the delayed decision was based upon evidence which was presented to the captain upon 5 different occasions by 3 different crewmembers, which evidence indicated that the fire was abating and that the smoke was subsiding, and that an emergency descent would not be required. -103NOTES ON N.T.S.B. REPORT AIR CANADA DC-9 C-FTLU CINCINNATI, JUNE 2, 1983 Submitted by: FlO C. Ouimet APPENDIX G ··The N.T.S.B. determines that the probable causes of the accident were a fire of undetermined origin, an underestimate of fire severity, and conflicting fire progress information provided to the captain. Contributing to severity of the accident was the flight crew's delayed decision to institute emergency descent.- So the criticism of the crew is contained in a time frame between 19:02:40 and 19:08:12 E.D.T., or the first 5 minutes and 32 seconds. Most of the explanation of that criticism is found on page 57 of the report. Paragraph 2, half way says: ··Neither he (flight attendant in charge) nor the first officer told the captain that they had not seen the fire and that they did not know exactly where it was or how lntensely lt was burning.··. J The board must have missed what is said by Mr. Benetti. on page 13 of Exhibit 12A, which is the transcript of the cockpit voice recorder. ··1 was able to discharge half of the COZ inside the washroom even though I could not see the source but it's definitely lnside the lavatory"· Those are the exact words of Mr. Benetti to captain Cameron. The report continues with paragraph 3: ··He (first officer) retreated the first time because he did not have smoke goggles with him...... In Exhibit 2A, first officer statement ··1 try to get tq the washroom but could not because of smoke··.· In the same N.T.S.B. report on page 3, paragraph 2 it says "The first officer said that he could not get to the aft lavatory because the smoke, which had migrated over the last three to four rows of seats, was too thick··. Also in Exhibit 12A, transcript of the voice recorder page 10, I say to the captain during my first visit to the cockpit ...•• 1 can't go back now, it's too heavy··. So I did not retreat and never said I did because I did not have my·smoke goggles, but because I could not physically walk to the washroom in that smoke. 1 thought •• •/2 APPENDIX G that, maybe, maybe using the smoke goggles would help me to get there with ~y tie around my nose and mouth.

[PROBABLE CAUSE] In this case, the captain first was told that there was a fire in the washroom, and he then was advised by his first officer that "it's too heavy, I think we'd better go down." Until that time, the captain did not have cause to believe that the fire would subside and he should have begun a descent immediately. Afterwards, he was misled by the crewmember reports. While the crewmembers should have realized that their comments could mislead and influence the captain's actions, the captain himself was ultimately responsible and he should have sought more positive verification that the fire had been extinguished. Further, the crewmember reports were only one element of information available to the captain. For example, the captain's comment "it's the motor" indicates that he did consider that the source of the fire might be electrical rather than paper burning in the wastebin. The captain should have known that an electrical fire which has progressed to the production of heavy smoke could be uncontained and out of control particularly when combined with his knowledge that circuit breakers had popped over 10 minutes earlier. This knowledge should have been sufficient to prompt the captain's immediate action. The Safety Board realizes that pilots must often make difficult, immediate decisions when confronted with abnormal or emergency situations. They often do not have the benefit of, or the time to weigh, all the facts. It is for this reason that flig'htcrews must be well trained to take the actions quickly that will be most effective in coping with the most severe and unexpected emergencies. The Safety Board had addressed its concerns in past safety recommendations regarding the need for effective cockpit resource management training to better prepare flight crewmembers to deal with complex, time-critical, decisionmaking situations, such as faced by this captain. Any indication of an in-flight fire justifies very rapid actions to land the airplane as quickly as possible. Therefore, the Board believes that the intent of the contributing factor statement must remain. However, the contributing factor statement has been revised since the final decision is the captain's alone. The Board considers the probable cause appropriate to emphasize the importance of an accurate and timely evaluation of the fire hazard by the flightcrew. ACCORDINGLY, -5APPENDIX H (a) The Petitioner's Petition for Reconsideration and modification of probable cause and findings of the airplane accident report on Air Canada Flight 797, McDonnell-Douglas DC-9-32, C-FTLU, Greater Cincinnati International Airport, Covington, Kentucky, June 2, 1983, is hereby granted in part. (b) The Board's original report is revised and a corrected report will be issued to the public which contains modifications to finding No.8 and the probable cause. (c) The probable cause is revised as follows: The National Transportation Safety Board determines that the probable causes of the accident were a fire of undetermined origin, an underestimate of fire severity, and misleading fire progress information provided to the captain.

[PROBABLE CAUSE] Had the crewmembers positively stated that the fire source had not been identified, the captain might have taken action sooner to descend in preparation for an emergency landing. With regard to the third issue, the Petitioner does not agree that contributing to the severity of the accident was the flightcrew's delayed decision to institute an emergency descent. While the Petitioner agrees that the time delay involved in commencing a descent contributed to the severity of the accident, the Petitioner believes that the cbntributing factor statement incorrectly implies that the captain's decision was different APPENDIX H -4from the decision that a more prudent captain would have made under similar circumstances. The Petitioner maintains that the captain did not have the benefit of the amount of time the Safety Board had to analyze the situation and make a decision. The Petitioner states, "... it is grossly unfair and unrealistic for the Safety Board to expect a flight crew to evaluate an incident in a dynamic situation, and reach the same conclusions in a few seconds that it took the Safety Board and all of its staff over a year to reach." Additionally, the Petitioner states that the Board could not even agree on the time when the crew's decision to descend should have been made during its "Sunshine" meeting on July 20, 1984. Therefore, the Petitioner requests that the contributing factor statement be changed to indicate that the delayed decision to descend was based on evidence provided by three different crewmembers on five different occasions which indicated that the smoke was abating and that an emergency descent would not be required. While it may appear to the Petitioner that the Safety Board's report unjustly focused on the flightcrew's delayed decision to land the airplane, the Board's conclusion was not made without a careful and lengthy review of the facts. The Board did come to an agreement regarding the time the flightcrew should have made a decision to descend1904:07-which is evident in the report, and there were no dissenting opinions filed by the Board's members after the "Sunshine" meeting. The Safety Board is of the opinion that the Petitioner's suggested change would not agree with the basic rationale of the Board's analysis of the flightcrew's handling of the fire situation. The Board believes that all pilots should be well a~are of the potential catastrophic nature of a cabin fire during flight. Therefore, the Board believes that any indication of a fire should prompt a flightcrew to begin immediate preparation for an expedient landing. In this case, the captain first was told that there was a fire in the washroom, and he then was advised by his first officer that "it's too heavy, I think we'd better go down." Until that time, the captain did not have cause to believe that the fire would subside and he should have begun a descent immediately. Afterwards, he was misled by the crewmember reports.

[CONCLUSIONS] These mounts are fused to fail at lesser loads than the wing, as a safety measure. The fuses held. In addition, the experts also believe that severe guet loads would cause the front spar to fail first, and that subsequently large sections of the wing would fall off the aircraft. Such gust loads would not be likely to tear off the high-frequency antenna and tip structure as separate pieces from the wingtip. Nonetheless, the NASA analysis did show that the mosi significant conclusion of this study is that turbulence alone can impose loads which exceed the ultimate design loads of the airplane structure. No "new" or generic problem surfaced during this analysis; however, the accident does serve as a reminder that turbulence associated with (hunderstorme can impose loads sufficient to cause failure of the pritury structural elements of modern transport aircraft. FINDINGS AND PLAUSIBLE HYPOTHESES~ (1) The aircraft was fueled with a mixture of JP-4 and Jet A fuels. (2) Lightning struck the aircraft an instant before an explosion. ™ (3) The first wreckage on the growd contained a considerable number of parts of the left wing outboard of No. 1 engine. (4) Damage to the wing in the area of the No. 1 fuel tank was the result of a low-order explosion. (5) The ullage or the No. 1 fuel tank contained a flammable mixture of fuel. (6) Pressures provided by the ignited fuel were suffictent to cause the damages. ‘ (7). Three fires occurred--in No. 2 tank, in No. 1 tank, and in the wingtip surge tank. (8) The crushing or collapsing of the fuel tube in No. 1 tank required an application of pressure only available from an explosion. (9) The pressure reguired to detach the stringers and skin from the wing were in the range of pressures developed by the explosion. (10) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) - 28 -- The first deposit of vreckage formed a pattern of light objects downwind and heavy cbjects upwind. ‘this pattern is not compatible with gusting or turbulent wind conditions, but is compatible with an explosion in calm or steady wind conditions. The high-frequency antenna and wingtip edge were snapped off the wing by inertial loads developed by an oscillating outer wing. The loosening of the stringer/plank unit from the wing destroyed the aft wing box of the wing. Extreme engine oscillations developed as a result of wing box damage. The loss of the rear box structure allowed the wing to twist torsionally and to deflect up and down about t’.e rear spar. The first objects along the flightpath were units from inside the No. 1 fuel tank. The three areas of fire within the left wing contained electrical devices, The higheet level of residual magnetic field was along the rear spar aft of the No. 1 tank.

[CONCLUSIONS] This evidence indficate: the followii g plausible sequence of events: The lightning first entered a forward part of the aircraft, perhaps on top of the cockpit, and exited from a static discharger on the left wingtip. As the aircraft continued forward, the flash hung on to the initial attachment point until the vertical fin reached the location where the forward part had originally beea. The flagh then reattached to th: vertical fin and continued to exit from the left wingtip. The lightning current's conductive patn to the static discharger at the ti» was through a bond strap along the trailing edge. Concentrationof curre.t at the riveted joint between this bond strap and a wing rib ceused melting an? release of molten metal and gasses; these were sufficiently conductive to cause the flash to reattach to this rfvet and to leave ine discharger. Before the apparent lightning strike, there were no unusual noises, or sounds of turbulence on the CVR recording. Immediately after an : ! ~ 24 - explosion was heard, there were sounds of objects bouncing around, some crashing sounds, and a discussion about loss of control. The fact that thc explosion occurred right after the thecrized lightning strike and in the wing which conducted the current suggests that a strike is plausible and was the cause. The strike current would have had to ignite the JP-4 fuel which was in the flammable range. Several possible places for the fuel to ignite were examined. The vent outlet--Fuel did not ignite at the fuel vent outlet. The lightning did not strike the outlet nor anywhere near it. Furthermore, the aircraft was descending, and air would have been flowing into the outlet and not out of it. Evidence indicated that the surge tank's protective system was operable but was not activated by a flame in the vent. Holes melted through tank skins--Neither lightning attachment points nor holes were found on any of the fuel tank skins. Thus, it was concluded that fuel did not ignite as a direct result of lightning attuching to the skin. . Electrical sparks at structural joints in fuel tank walls and skins-- The possibility of ignition by this cause was remote since the structure was so massive. Access doors and filler caps--The access doors and filler caps are not located in probable lightning-strike zones on the aircraft; no strike evidence was found on them; and they were coated with conductive material to guard against the very remote possibility thar they could be struck, Sparking at fuel quantity measurement devices as a result of induced voltsges--Fuel did not ignite at the overfill compensator probe located in the wingtip.

[CONCLUSIONS] The level of residual magnetic field strength in this area of the wing was indicative of high currents. Lightning certification tests indicatea that this area about the rear spar was a lightning attachment zone. A domestic carrier experienced electrical failures in several motors of the fuel valves after the aircraft was struck by lightning. Lightning currents penetrated the motor circuits and short-circuited electrical filters which disabled the motors. The evidence (i) that the explosion in the No. 1 tank occurred in the immediate area of a motor-driven fuel valve, (2) that the motor was nevet recovered, (3) that a high level of residual magnetic field still existed in the ferrous material at this area, (4) that certification tests showed this area to be a likely lightning-attachment point, (5) that the Lightning strike is known to have disabled the motors on other aircraft, and (6) that no other possible ignition source could be determined provides a foundation for an hypothesis that the tank explosion could have been ignited by a spark at this motor-driven valve. Similar systems on other aircraft--Assuming that a lightning strike can generate a source of ignition to fuel vapors, aircraft fuel explosions could occur more frequently. However, events must combine simultaneously to create the explosion, and this combination would occur rarely. In this case, the events were: (1) An intermictently conductive path which closed and opened an electrical loop, (2) a lightning-induced current of sufficient intensity flowed in this path and formed & spark, and (3) a flammable vapor surrounded this spark. Possibly this combination of events has occurred a number of times before, in the following accideats: * Milan, Italy (Constellation) * Elkton, Maryland (B-707) * Madrid, Spain (USAF KC~135}} * KSC, Florida (USAF F-4) — * Pacallpa, Peru (L-188) “Ta phage Nee RE rm er wom et fn tim Sat it A EN PG et i AI el allt I NR AI en RET A ote me te ae i te Rm me NE A Te ee ce me ae ee . : ‘ Evidence of a lightning strike to a w'ng followed by an explosion in the same wing exists in each of these cases, yet no specific lightningrelated cause, such as ignition at a vent outlet, was found. Structural Overload due to Gust Penetration or Turbulence The most liicely alternative to destruction of the wing by Lightning and explosion is its destruction by turbulence. This alternative gains credibility if much of the evidence is interpceted accordingly. The CVR tape shows that violent weather conditions existed along the flightprth. The aircraft was vectored around oie thunderstorm but : another lay aheac. The captain's remark that the weather ahead "will tesr us apart’ if entered, and another crewmember's remark that "we're in the soup" after the captain's statement could indicate that the aircraft had entered the thunderstorm.

[CONCLUSIONS] The failure began at WS 1140 and progressed inboard and outboard from that position. The upper skir. shear tie attachments at the nacelle support rib and the flap track support rib fractured in bending because of the continued upward movement of the upper surfece. The upper surface stiffener ties to the nacelle rib separated becavse cf the outboard movement of the nacelle rib; the outboard movement was caused by the overpressure on the inboard side. When the upper wing skin panel which was attached to the mid and rear spars separated, the aeroelastic properties of the wing, and especially the outboard section of wing were altered drastically. The stiffness vf the No. 1 engine mount waa greatly reduced in the pitch axis by the loosening of the skin and the resultant loss of wing box integrity. The loss of structural integrity of the wing box permitted increased torsional deflection of the wing. The outer wing began to oscillate, and lateral loads were generated by the vibrating engine. The oscillations developed inertial loads on the higtfrequency antenna and outer tip and caused them to s:parate. The changing aerodynamic Joad on the outer wing and the lateral forces generated by the oscillating engine caused compression faflures in the upper skin above the deflecting rear spar. This compression fracture pre - gressed over the whole epan of upper wing skin. The front bex maintained the structural ittegrity of the forward wing until oscillations of the outer wing (torsional bending) and engine-induced lateral loads caused its destruction. Lightning As An Ignition Source Based on the hypothesis that an explosion occurred in the No. 1 fuel tank, the lightning strike became a plausible source of ignition. During its descent into Madrid and chortly after descending theough 10,000 ft, the aircraft was apparently struck by lightning. The following observaticns and events support this hypothesis: (1) The cockpit statement of too much “activity" in front and the request for a vector around 130 seconds before the end of the CVR recording; (2) the cockpit discussion about an active "CB" in front 86 seconds before the end of the CVR; (3) the cockpit remark about being "in the soup;" (4) the audible sound and electri:21 transients on the CVR recording 52 seconds before the end of the recording; {{5) eyewitness reports: %ne who said lightning struck the aircraft "midway between the (No. 1) engine and the wingtiy,'' and another who reported seeing the aircraft get struck with lightning that “wouldn't go away" and the aircraft "flying off in flames;" surface weather reports of cumulus clouds or thunderstorm activity in the area; and (7) the physical avidence of lightning attach points on the wreckage.

[CONCLUSIONS] CONCLUSIONS Af av analyzing all of the available evidence, it is cencluded that. the nosc probable sequence of events which culminated with multiple struccurai fafiures and separation of the wing began with an ignitior. of the fuel vsepors in the io. 1 fue) tank. The damage te the structure in the area of the tank provided positive indications of an explosion. The possibility that the explosion was a secondary result of structural failure caused by excessive ceroJynamic forces developed during high velocity gusts and turbuJerce cannot %e completely dismissed; however, the evidence and the probebilities of an aircraft's encountering these unique environmental conditions make this hypothesis less supportable. Ignition of i*uel Vapor in No. 1 Fuel Tank By accepting the hypothesis that the explosion {{in the No. 1 tank was the first destructive event, the varfous wing failures car. be explained as follows: The explosion failed the fasteners ttat held the stringers to the ribs and the skin to the spars; the integrity of the aft wing box wae lost as a result, which greatly redu'ed the torsional strength of the wing; and support of the No. 1 engine in the pitch plane also was lost. The fact that explosive forces could be developed in the tank verified that the wing skin forming the top of the tank was whole before the explosion. The probable sequence of wing destruction follows: (1) Overpressure was generated in the No. 1 fuel tank as a result of ignited fuel vapors. The location of this overpressure was the aft outboard corner of the tank adjacent to the closure rib or the nacelle rib. Because of the overpressure, the upper skin panel, including stiffeners, pulled loose from ribs inboard of the nacelle rib and afc of the mnidspar. The stringer to rib fasteners separated by contined tension and shear, which resulted from the overpressur. cnd subsequent inboard dispiacement of ribs at WS 1140 and WS 1168. LO es La ARON eh SESE abha a niane Neue aca bt Meg ae alee foe nuk NK ene Se 6 PS in cA ned ag Om WAL HhiENS By isin eae: alte udann Rete Ai ions © Pe eet eer rt rar Se ero 2c a i Let seh Reeling Aas AOL hs OR. anahaiaeCiintY ace Senne Naame (I Mee ieee The inboard displacement of the ribs ruptured the rib attachments to the lower surface and spars, and the ribs becane detached, The upper skin panel billowed upward because of the explosion until bending fractures occurred at the mid and rear spars and the fasteners were sheared. The failure began at WS 1140 and progressed inboard and outboard from that position. The upper skir. shear tie attachments at the nacelle support rib and the flap track support rib fractured in bending because of the continued upward movement of the upper surfece.

[CONCLUSIONS] Reserve Tank Report - G. Economy, USAF Hypothesis - Explosion, Grabowski, Feawal, May 30, 1976 Surge Tank and Vent Soot Patterus, Supplement, Horeff, FAA Fuel System Components, Damages, T. Curran, FAA, SEA Hypothesis - Fire Initiation, A. Voeller, AAL, June 30, 1976 Electrical Devices, Wiring and Bonding, Schwind and Zion, ALPA and FAA, August 13, 1976 Drawing, Electrical Devices - Jobe, Boeing, July 1, 1976 STP Components Test - Andrews, Fenwal, January 24, 1976 Soot, Fire and Heat - Discussion, Binding, DACO Tests: Boost and Jettison Pumps, Boeing, September 28, 1976 Charts: Estimated Flightpath and Distribution, Boeing Cormpcusator Capacity Measuremente, NAFEC Labs, June 31, 197€ VOR Antenna and Wingtip lightning Damages and Drawings, Zinn, PAA Wingtip Structure Damage . Hypothesis, Williams, Pan Am, Jpne 2, 1976 VAL, 747, Incident Report and Lightning Data, June 7, 1976 Compensator, Physical Exam, Drawings, and Notes, NTSB, Honeywell, FAA, GE Residue Auclyses, Septemter 8, 1976, Boeing, FBI Fuel System Component Exams June 30, 1976, and September 28, 1976 Fire and Exploston Study, Grabowski, Fenwal, November 12, 1976 Compensator Breakdown Tests, 6/76, Plumer, GE Static Discharge Reports, B-747, Lavers, Air Canada, June 4, 1976 Wing Structure Factual Report, November 30, 1976, Williams, Pan Am Fuel Line Collapse Test, October 28, 1976, Boeing -~ 34 - Wichita Modification Plan, 747, Document No. D3-8168 -50, Boeing, August 28, J975 Upper Wing Skin Exam, NTSP, Marx, June 30, 1976 Damage Report, Septcider 20, 1976, Battelle Summary and Discussion, Fewtrell, October 19, 1976, Lockheed Califor-.a Wing Skin Group Report, August 1976, J.

[CONCLUSIONS] D3-8168 -50, Boeing, August 28, J975 Upper Wing Skin Exam, NTSP, Marx, June 30, 1976 Damage Report, Septcider 20, 1976, Battelle Summary and Discussion, Fewtrell, October 19, 1976, Lockheed Califor-.a Wing Skin Group Report, August 1976, J. Houck, NTSB Overpcessure for Skin Loss, Table, Boeing, Marct. 15, 1977 Danage Report, Fuel Overpressure Invident, Boeing Mid Spar Fuse Pins, Boeing Flame Propagation Rates, Boeing Electrical Transients, CVR Analysis, July 19, 1976, Turner, NTSB P & W Letter, re engine Vibrations, dated May 20, 1971 CVR Factual Repovt, May 21, 1976, Turner, NISB Sparking Tests, Fuel Vent Couplings, December 2, 1976, Plumer, GE Hypothesis and Recommendations, March 31, 1977, Plumer, GE Fuel System Drawing and Notes, Plumer, GE CVR Transient Comments August 17, 1976, Plumer, GE Boeing Conclusions, Ex iosion etc. and Photos, November 2, 1976, Boeing Tank Access Doors, Wirg Plank Sample, Report, February 11, 1977, Boeing Load Transfer - Spars and Stringers to Skin, November 1976, Boeing Stringer Loads, November 1976, Boeing, Pottinger Lower Skin Marks and Nacelle Punctures Report, December 20, 1976, Boeing IIAP VOR Antenna Tests, February 11, 1977, Boeing Accident Report, February 1977, Lanoway, NTSB Damage Report, Final, Janvary 31, 19/77, Battelle Additional Comments, May 1977, Fewtrell, Lockheed California FDR Report, Carroll, December 7, 1976, N1'SB Lightning and Explosion Examination, Hacker, NASA (ret), February 3, 1977 Alert Service Bulletin, P & W, No. 3108, March 1972 Telegram: All JT9D Operators, P & W, January and February 1972 Wing/Engine Vibration, B~747, American Airlines, August 27, 1976 Fuel System Analys’s, Failure Analysis, Letter, December 3, 1976, Boeing Fuel System - Ignition Protection, Fecember 9, 1968, Boeing AD, Boeing, June 20, 1977 KC~135 Accident Report, June 3, 1971, Tarrejon, Spain, USAF Structural Analysis, January 27, 1977, Wade, Lockheed Georgia Electrical Inspection Report, September 30, 1976, Boeing Wing Failure Analysis, Report. to FAA from Boeing, October 5, 1976 KC-135Q Accident Summary, USAF Wing Drawing of Fire Heat Soot, Bob Ray, Lockheed California, Novemcer ll, 1976 Wing Fir: Study and Analysis, Rinding, DACO, November 24, 1976 Strike Evidence on CVR, Plumer, GE, December 1976‘ Compensator Microscopic Exam, Plumer, GE, November 1976 Wiggins Coupling Arcing Exam, Plumer, GE March 1977 Chart, CVR with Flightpath, Bruggink Lightning Exam, Wiggins Coupling, Magnetic Fields, Mangold, USAF, February 7, i977 Autopiloc Hardcover and Flammability of JP~4, Boeing, July 8, 1977 Hardovers and Nsciliatory Control Loads, D6-3087/0-3, B-747~100, Boeing Structural Design Loads, D6-13248-3, B-747, Bowing Tube Collapse Test Photos, August 1977, Battelle Dynamic Loads Group Report, NASA, Houbolt, April 1977 CVR Update, May 1977 Turner, NTSB Special Studies, CVR, Turner, May 1977, NTSB Metallurgist Analysis Report, April 27, 1977, Houck, NTSB Lightning Protection B-747, Document D6-13487, Boeing STP Certification, FAA/Boeing, November 25, 1969 Calculated Pressure and Strain Energy, March 15, 1977, Boeing Structural Design Loads, B-747, Document No.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] It could not be determined to what cxtent the pilot followed the recommended procedures for an engine failure or fire in-flight. ‘The indication that the left propeller was feathered at impact was not necessarily the result of the manipulation of cockpit controls; in case of loss of engine oil pressure the propeller would feather regardless of pilot action, if the loss occurred at sufficient pm. The focl and oil shutoff valves behind the firewall were cable controlled; duc to the stretching and separation of these cables during the airframe breakup, the postcrash position of these valves would be an unreliable indicator of pilot actions, The switch thay controls the boost pump in the left inboard fuel cell was not recovered, A fire-induced failure in the pressure live to the fuel selector would result in fuel spillage regardless of selector position, With regard to the propagation of the fire from the engine compartment into the wheel well, the available evidence suggests that the fire burned through the left engine cowling, and progressed around the firewall, and then burned through the wheel-well skin immediately below the right augmenter tube. In this manner the fire would have entered the wheel well, below the fuel selector valve installed on the rear of the firewall. The burning of one or more of the aluminum fuel lines in this area would lead to an intense, uncontrollable wheel-well fire. The direct exposure of the unprotected wingspar caps to the fire in the wheel well led quickly to hea weakening of these load-carrying structures. The most intense fire occurred at the rear spar location; this spar was probably the first to lose its structural integrity, The subsequent increase in the tensile loads on the heated lower forward spar cap resulted in an upward overload failure of the left wing. 2.2 Conclusions (a) Findings 1, The aircraft was certificated in accordance with existing regulations. Improper maintenance procedures were utilized during the installation of the cylinders on the left engine. The pilot was certificated and qualified for the flight. Fire in the left engine commenced with the separation of No, 5 cylinder :ssembly. Uncontrolled fire progressed from the engine nacelle into the wheel well. The left wingspar cap was weakened by heat and eventually caused the in-flight separation oF the lef wing, The aircraft structure of the Beech Model 65 lacks adequate fire protection. (b) Probable Cause The Nationa! Transportation Safety Board determines that the probable cause of this accident was an uncontrolled fire in the left wheel well which resulted in loss of structural integrity of the left wingspars.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] The melting of the aft bracket that attaches the first stage of the augmenter assenibly to the engine support, as well as the burning of the adjacent cowling, the fucl lines to the fuel-flow transmitter, and the generator harness, indicated that a severe fire was burning in the right, lower section of the engine nacelle. Although the pilot would have been aware of an eigine problem when the No. 5 cylinder ceased to operate or caused vibration, it is unlikely that he would have been aware of the initial engine ce Oe Ale NAb Cal MINE RIE A Sy mtg . berg . Sob abigd vate gra BANE A eke iteneS hcota yb ne lib Bethel SOAR ED IMT ONE Le Se Salih MRR END, Aiea td RUTOLIR WER MOK NORMAL NI RO INE ain TY AnH ee comet AAG pater FMM MUM iAH MBA OR we le eT a ee Se caaitier’ ta Sch ame ttle ae RS oh edhe itn dartacea Sogn Epedai dows 4 Aas et ce ene NAOT HAR TSS ORIG fire, Most of the smoke and flames would have been forced outside, under the wing, through the exhaust augmencter tubes. These indications of fire would not have been readily apparene frora che cockpit until the right side of the left engine cowl or the upper wingskin above the wheel well burned through. Based on witnesses’ report of unusual engine sounds, and the mutilated condition of the No. 5 piston, it appears that the pilot did not immediately secure the left engine when the No. 5 cylinder became inactive. Despite vibrations, some power would still have been available, which might have prompted the pilot to attempt a landing on the cleared, hard-surface runway at Fairfield. The left turn described by the witnesses scoms to support this speculation, although this also happened to be the direction of predominant asymmetrical thrust. It could not be determined to what cxtent the pilot followed the recommended procedures for an engine failure or fire in-flight. ‘The indication that the left propeller was feathered at impact was not necessarily the result of the manipulation of cockpit controls; in case of loss of engine oil pressure the propeller would feather regardless of pilot action, if the loss occurred at sufficient pm.

[(b) PROBABLE CAUSE] The left wingspar cap was weakened by heat and eventually caused the in-flight separation oF the lef wing, The aircraft structure of the Beech Model 65 lacks adequate fire protection. (b) Probable Cause The Nationa! Transportation Safety Board determines that the probable cause of this accident was an uncontrolled fire in the left wheel well which resulted in loss of structural integrity of the left wingspars. The whecl-well fire resulted from an uncontained fire in the engine - ese sete Ras Raed bein ek Se an ee AER TUNER 6 WARP ASC cake wah emaelneana. ibe ameett emnecneryetuaie pep necnerener ernie oae ee Me EA ED Lett WEEN AON ORES NEED Stites ters nee atietate te SOO RAM NTRS BRR ETON AI Rp ERM il compartment which, in turn, was initiated by separation of one of the engine cylinders due to the use of improper maintenance procedures. Excessi-e working hours may have contributed to the oversight by the maintenance personnel involved, ous deviation from the design premise that an aircraft, should be able to tolerate an cngine failure, atu even an engine fire, without immediately affecting the aircraft’s structural integrity. The Board's Safety Recommendation A-72-21 through 24, issued March 3, 1972, and addressed to the FAA Administrator was an expression of that concern. The Administrator responded to these safety recommendations on July 5, 1972, The response indicated full compliance with the Board’s recommendations. (See Appendix E.) 3. RECOMMENDATIONS This accident, as well as a similar one in Australia, a month earlier, demonstrates a seri- BY THE NATIONAL TRANSPORTATION SAFETY BOARD; fsf JOHN H. REED Saget ercoeyantenrncncien ARAN NTP ARE RAE od RRNA AMERY Fe rine, BEANE SATE HEEL P AAR AD TIRED = MMR AIEEE ARE E Chairman /s) FRANCIS H. McADAMS eee reg bate Member “ee I se neem AARNE ann Ries oneal nie NNER AE Is}} 'tSABEL A. BURGESS Meisber PAO AE OREN CHR MOIR Re HEL RT OURO KEEN ATA OIE /sf WILLIAM R. HALEY _. Member Louis M. Thayer, Member, was absent, not voting. August 30, 1972. se auasasdae: 55 henge . SRRIR RESTA RUSTE PAERY TICS EET TRS SS SeLmMeLkRY eke Mamma eT Ate On mH mT NUE NA HAs PRES NER MMT REE ORE I RO AIO AO BO OMRON SES ed ahis SSUES RRS eat AGRE had age Waitt Biers aan as aes APPUNDIX A INVESTIGATION AND HEARING

[ANALYSIS AND CONCLUSIONS > ANALYSIS] This could occur cnly if the nuts were not installed, or if they were installed but not tightened. The engine manufacturer reported that untorqued nuts have been observed to back off the studs in approximately 10 minutes engine operating time. The three recovered lower studs failed as a result of alteriating bending overloads produced by the cylinder pivoting about these studs with the nuts probably still in place. When these studs eventually failed, the cylinder was released and moved away from the crankcase, at the same time displacing the exhaust stacks away from the augmenter tube and probably against the cowl door. It is apparert that the engine manufacturer's cylinder teplacument procedures were not followed. Had the prescribed tightening sequence been followed, it is highly unlikely that any nuts would have been omitted or left untightened. It should be noted, however, that there are convincing reasons which attribute the improper maintenance procedures in this case to fatigue of the mechanics involved, rather than to their carelessness or incompetence. In order to have the aircraft ready for the Sunday morning departure, the two mechanics worked approximately 27 hours on the preceding Friday and Saturday. ‘They finished their task around midnight, Saturday. The error potential implied in such a situation is a direct reflection on the quality control of the airline’s management maintenance program, The reason for the left-engine fire was directly related to separation of the No. 5 cylinder. This separation allowed engine oil to be pumped and spilled into the engine compartment in the immediate vicinity of the right augmenter tube. In addition, the No, 5 intake manifold pipe probably separated from the central induction system, thereby releasing a combustible fuel-air mixture, Ignition could easily have been caused by the exhaust stacks of the Nos. 1 and 3 cylinders or by a broken ignition wire from the No, 5 cylinder. The melting of the aft bracket that attaches the first stage of the augmenter assenibly to the engine support, as well as the burning of the adjacent cowling, the fucl lines to the fuel-flow transmitter, and the generator harness, indicated that a severe fire was burning in the right, lower section of the engine nacelle.

[ANALYSIS] Details of the examination of each of these 48 NTSB Aircraft Accident Report 49 sources can be found in appendix C. However, none of these were proved conclusively to be a source of the combustible fluid. The NTSB concludes that a combustible fluid, such as oil, hydraulic fluid, or fuel, was available in the engine; however, fire damage precluded the determination of the specific source of the combustible fluid. 2.4 Flight Crew Performance As noted, during the flight, the pilots encountered an uncommanded opening of the ATSV followed by indications of an engine fire. During the brief flight, the pilots also encountered several other abnormal events, including electrical and hydraulic system anomalies and the nose landing gear’s failure to extend. The investigation revealed that the flight crew did not perform several of the appropriate checklists and interrupted an emergency fire-related checklist. This section discusses in detail the pilots’ responses to these anomalies and the possible reasons for them. According to CVR evidence, about 74 seconds after the takeoff roll began and while the airplane was climbing through about 1,500 feet mean sea level, the first officer detected the illumination of the ATSV-Open light for the left engine. About 1 minute later (1313:55), the Left Engine Fire aural warning sounded, and the Left Engine Fire warning light illuminated. A review of the CVR transcript revealed no evidence that the flight crew performed any of the L or R Start Valve Open checklist items during the 53-second period from the detection of the ATSV-Open light to the onset of the Left Engine Fire warning.59 This section details the pilots’ actions in response to both abnormal and emergency alerts and some of the possible reasons for and the effects of these actions on the events that occurred both in flight and after landing. Section 2.5 considers additional human factor issues that were identified during the investigation. 2.4.1 Detection of and Response to the Left Engine ATSV-Open Light The proper operation of the ATSV-Open light system could not be completely verified during postaccident inspections because the wiring harness within the nacelle sustained too much fire damage to confirm electrical signal continuity. The time that the ATSV-Open light illuminated could not be determined through aircraft inspection or CVR or FDR information. During the Before Takeoff checklist, the pilots checked the status of the annunciator panel, and they made no comments about the illumination of the ATSV-Open light at that time. 59 In postlanding statements recorded by the CVR and during postaccident interviews, the pilots indicated that they thought that only a few seconds elapsed between the illumination of the ATSV-Open light and the Engine Fire warning. NTSB Aircraft Accident Report 50 No direct evidence exists to determine exactly when the light illuminated during the 74-second period between the application of takeoff power and the first officer’s detection of the ATSV-Open light at 1313:02.

[CONCLUSIONS > FINDINGS] The pilots might not have immediately detected the air turbine starter valve (ATSV)-Open light illumination because of its location, static appearance, and color, and, once they detected the light, the pilots did not immediately respond to it because an open ATSV was considered an abnormal situation that did not require immediate action and they were involved in air traffic control communications and airplane configuration changes. 12. Coupling the air turbine starter valve (ATSV)-Open light with the Master Caution system might increase pilots’ ability to detect the presence of an abnormal ATSV condition; however, unintended consequences, such as aborted takeoffs, may occur and more work needs to be done to determine whether the Federal Aviation Administration should mandate the modification of the ATSV-Open light in the MD-80 fleet. 13. The pilots failed to properly allocate tasks, including checklist execution and radio communications, and they did not effectively manage their workload; this adversely affected their ability to conduct essential cockpit tasks, such as completing appropriate checklists. 14. No preexisting indicators in the pilots’ training or performance histories were found that could explain their poor performance during the accident flight. 15. The pilots’ interruption of the emergency Engine Fire/Damage/Separation checklist at a critical point prolonged the fire and led to additional problems, including the loss of hydraulic pressure, which caused the nose landing gear to fail to extend. 16. Given the airplane’s altitude and the lack of time to prepare for a nose landing gear up landing, the captain’s decision to go around was a reasonable choice. 17. The captain’s decision not to conduct an emergency evacuation after the airplane landed was in accordance with company guidance and was appropriate because the fire was not severe and aircraft rescue and firefighting personnel were actively responding to the residual fire. 18. The incident commander’s decision to deplane the passengers after fuel spilled out of the engine area was prudent. 19. The first officer did not have a clear understanding of the relationship between the pneumatic crossfeed handle and the engine fire handle, most likely because of inadequate company 67 NTSB Aircraft Accident Report guidance and training on the issue; this resulted in the first officer inadvertently reintroducing fuel to the left engine, creating potential unnecessary risk of fire. 20. Improved pilot training methods for responding to multiple systems failures, competing task demands, and increased workload would help pilots develop the skills and decision-making abilities needed during both single and multiple abnormal and emergency situations. 21. The casual atmosphere in the cockpit before takeoff affected and set a precedent for the pilots’ responses to the situations in flight and after landing, eroded the margins of safety provided by the standard operating procedures and checklists, and increased the risk to passengers and crew. 22. Operational procedures requiring that an airplane be configured for an evacuation when it is stopped away from the gate after a significant event would help expedite an emergency evacuation if one became necessary. 23. During the emergency situation, the flight attendants did not relay potentially pertinent information to the captain in accordance with company guidance and training. 24.

[ANALYSIS] Further, the CVR did 54 NTSB Aircraft Accident Report 55 not record the pilots completing the Engine/Fire/Damage/Separation checklist. Specifically, they did not start the APU, turn the fuel crossfeed on, or set the hydraulic system. About 1319:59, the flight crew attempted to extend the landing gear; however, the gear indicator lights did not provide valid information about gear status. Neither pilot detected the absence of the normal increase in cockpit noise associated with the extension of the nose landing gear67 or the loss of hydraulic pressure on the right side, which they might have detected if they had taken time to distribute their tasks and to become more deliberate and focused. Noticing these conditions might have helped them to detect the nose landing gear problem earlier. However, the NTSB recognizes that the detection of this anomaly would have been more difficult because of the errant cockpit indications caused by the fire-induced electrical problems. As events continued, rushing through and skipping tasks continued to cause problems for the pilots. The first officer asked ATC to verify whether the landing gear had extended. ATC responded that the main gear appeared to be down but that the nose gear did not appear to be down. If the pilots had continued performing the Engine Fire/Damage/Separation checklist, they would have reached the item to set the hydraulic systems, which would have provided them with an opportunity to notice the hydraulic anomalies that were being caused by the engine fire. If the pilots had detected the hydraulic anomalies, the pilots should have completed the checklist and then initiated the Left or Right Hydraulic Pressure Low or Hydraulic Quantity Low or Decreasing checklist. The first step on both checklists was to turn off the hydraulic transfer pump switch, which would have allowed the right hydraulic system to pressurize and enabled the nose gear to fully extend. The CVR did not record the flight crew performing either checklist during the flight. According to American Airlines, pilots were trained to check hydraulic pressure and quantity indications before responding to the hydraulic system-related item on the Engine Fire/Damage/Separation checklist. The NTSB concludes that the pilots’ interruption of the emergency Engine Fire/Damage/Separation checklist at a critical point prolonged the fire and led to additional problems, including the loss of hydraulic pressure, which caused the nose landing gear to fail to extend. 2.4.3 Decision to Go Around After ATC verified that the nose landing gear had not extended, the captain decided to execute a go-around. During postaccident interviews, the captain stated that, by the time he was aware that the nose landing gear had not extended, the airplane was too low and close to the airport to manually extend the nose landing gear and that, even though fire indications were still alerting in the cockpit, he no longer trusted the annunciators because of the airplane’s electrical anomalies. The captain also stated that he did not want to attempt a landing without the nose gear 67 As demonstrated in other DC-9 airplane wheels-up landing accidents, pilots can fail to detect the absence of increased cockpit noise in high-workload conditions.

[ANALYSIS] During postaccident interviews, the captain stated that ARFF personnel gave him the sense that they had the situation under control. Further, both pilots stated that, because ARFF personnel were actively applying fire-extinguishing agent to the residual fire, they concluded that it was safer for the passengers to remain on the airplane. About 1337, ARFF personnel reported that the fire was extinguished. The NTSB concludes that the captain’s decision not to conduct an emergency evacuation after the airplane landed was in accordance with company guidance and was appropriate because the fire was not severe and ARFF personnel were actively responding to the residual fire. 56 NTSB Aircraft Accident Report 2.4.5 Retraction of the Fire Handle While waiting for the tug to tow the airplane to the terminal, the first officer opened the left pneumatic crossfeed valve to provide air to the cabin. Shortly thereafter, ARFF personnel informed the pilots that fuel had spilled out of the engine area onto the ground, and the first officer pulled the left engine shutoff valve. Subsequently, the IC recommended to the captain that the passengers be deplaned as a safety precaution, and the captain complied. All of the passengers deplaned without injuries. The NTSB concludes that the IC’s decision to deplane the passengers after fuel spilled out of the engine area was prudent. According to the Boeing MD-80 FCOM, opening the pneumatic crossfeed valve on the MD-82 airplane causes the associated engine fire handle to retract. In this case, opening the pneumatic crossfeed valve caused the left engine fire handle to retract and fuel to be reintroduced to the left engine area. Although the American Airlines AOM indicated that the fire handle was mechanically linked to the pneumatic crossfeed lever and pilots are told during training that pulling the fire handle will mechanically shut off the pneumatic crossfeed valve, American Airlines’ manuals and training did not indicate that opening the crossfeed valve causes the fire handle to retract, reversing the shutoff of fuel. The NTSB concludes that the first officer did not have a clear understanding of the relationship between the pneumatic crossfeed handle and the engine fire handle, most likely because of inadequate company guidance and training on the issue; this resulted in the first officer inadvertently reintroducing fuel to the left engine, creating potential unnecessary risk of fire. As a result of this accident, the FAA issued SAFO 08018, which explained the fire handle characteristics of DC-9, MD-80, and MD-90 series airplanes and the relationship of the engine fire handle and the pneumatic crossfeed valve. The SAFO recommended that operators review their training and operating manuals to ensure that the design and interrelationship of the systems affected by the fire handle are adequately explained. It further recommended that they add a caution to their checklists stating that opening the crossfeed handle will retract the fire handle and potentially reintroduce fuel to a fire.

[ANALYSIS] About 1317:26, the CVR recorded the first officer stating, “I can’t even shut it off,” referring to the difficulty he was experiencing moving the fuel lever to the OFF position, which was the third item on the checklist. About 1317:36, the first officer read the next item on the checklist, which directed the nonflying pilot to pull the engine fire handle on the captain’s command. The CVR did not record the captain confirming that the first officer had pulled the fire handle; however, about 1317:52, the first officer advised the captain that both fire agents had been discharged, which is the next item on the checklist and could only have been accomplished if the fire handle had been pulled. Even though the checklist had not been completed, the first officer then announced that he was going to set the landing speed settings. CVR evidence indicated that the first officer was subsequently distracted as he tried several times to shut the cockpit door, which kept opening as a result of electrical problems, and responded to several ATC transmissions. The NTSB is concerned about how the pilots managed and prioritized tasks during this phase of the flight and is especially concerned about the flight crew’s interruption of the emergency Engine Fire/Damage/Separation checklist. During the 2 minutes that elapsed from the time that the first officer initially started the checklist to the time he resumed it, the first officer engaged in radio communications for about 63 seconds, and the captain engaged in the flight attendant briefing for about 31 seconds. About 80 percent of the pilots’ delay in performing the third item on the checklist, which would have shut down the fuel supply to the engine and alleviated the fire, was caused by their performance of tasks unrelated to shutting off fuel to the engine and stabilizing the fire. To minimize the severity of the fire, the pilots should have continued conducting, without interruption, the Engine Fire/Damage/Separation checklist up to, at a minimum, completing the following critical items: shutting off the fuel to the affected engine, pulling the associated engine fire handle, and discharging the fire agent. By delaying the performance of these critical items, the pilots exposed the passengers, cabin crew, and airplane to unnecessary risks. American Airlines stated that its pilots were trained in the simulator to complete the checklist to at least the point where fire agent was discharged before deviating from the checklist. Pilots are expected to know the possible consequences of interrupting or deviating from an emergency or abnormal checklist, especially if a fire is involved. Further, the CVR did 54 NTSB Aircraft Accident Report 55 not record the pilots completing the Engine/Fire/Damage/Separation checklist. Specifically, they did not start the APU, turn the fuel crossfeed on, or set the hydraulic system. About 1319:59, the flight crew attempted to extend the landing gear; however, the gear indicator lights did not provide valid information about gear status.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] Also, determination of the aerodynamic effect of ice on 4 specific airfoil is not a requirement in the certification of the aircraft. Even if this were a requirement, it would be virtually impossible to determine the aerodynamic penalties of the various shapes and textures of ice that may be accumulated in flight. Studies conducted by the National advisory Commtt tee for Aeronautics and later by the National Aeronautics and Space Administration, Féederal Aviation Administration publications, and the textbooks previously cited, have confirmed that airfoil icing does result in aerodynamic penalties and that, {{n general, these penalties result in higher etall speeds and lower stall angles of attack, In view of the foregoing, the Board conctudes that airfoil {{cing existed to such an extent that substantial aerodynamic penalties were imposed, These penalties, while not precluding the aircraft from becoming airborne and briefly establishing a positive rate of climb in ground effect, resulted in a stall as soon as it lost the advantage of ground effect, Accordingly, except for ground effect, the aircraft probably would not have become airborne under the existing circumstances, I - The experience level of the flightcrew in this aodel aircraft was minimal, However, the Board believes that the relatively low level of experience should not have been a factor in the accident, since icing effect on the airfoils of an aircraft is basic knowledge required of all certificated pilots, While ice is more critical on a thin swept-wing aircraft, the same basic precautions are applicable to all aircraft regardless of airfoil design. The Board concluded that the validity of information provided by the speed command system and the stall warning. system with an iced airfoil is questionable, A stall can occur with an iced airfoil at a lower angle of attack than that programaed for the stall warning system and, under extreme icing conditions, the validity of the information provided by the speed command system becones questionable, However, proper use of the airfoil antisice system and proper deicing of aircraft prior tc departure would éliminate this opera-. tional consideration, I = 28 « The popping and torching of the engine near the time of the stall resulted from engine compressor stalls caused by the crew's manipulation of the throttles during their attempted recovery and subsequent abandonment of the takeoff, and disturbed engine inlet airflow resulting from the stabled attitude of the aircraft. This phenomenon has beén observed in other jet aircraft accidents.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] The Board can only conclude that the crew failed to check the weight and balance form, they read it erroneously, or they were not thoroughly knowledgeable of the power requitements of this aircraft, Although the crew had satisfactorily completed the prescribed training in this model aircraft, their actions on this flight did not demonstrate a thorough knowledge of the thrust requirements or of the use of the anti-ice systems, While {{t cannot be substantiated that the use of el power would have prevented this accident, the increased acceleration would have increased the probability of a successful takeoff, Although tower personnel, a number of passengers, and ground witnesses thought that the takeoff acceleration of this aircraft was slow, 4 comparison of the acceleration, as determined from the flight data recorder with the computed expected acceleration, disclosed that the acceleration was very near the computed expected acceleration for the power setting selected, The setting actually used by the crew for this takeoff, while not in accordance with performance requirements, was adequate to permit a successful takeoff on Runway 35, with clean airfoils, The increased power required for the existing gross weight is stipulated to meet the single-engine climb requirements in the event of an engine failure following takeoff. _ The flightcrew confirmed that lift-off did occur and that a positive rate of climb was noted prior to theloss of control, The review of the flight data recorder readout indicated that lift-off occurred at near the computed perfor.sance lift-off speed. This might seem to question the influence of the aerodynamic effect of I the ice adhering to the airfoils of the aircraft during this takeoff, » 27 « However, studies have confirmed that the increased lift obtained frou ground effect may permit an aircraft with iced airfoils to become airborne and subsequently be unable to maintain fiight when entering the area of reduced Lift leaving ground effect, The accurate weight, size,and shape of the ice adhering to this aircraft at the time loss of control was experienced could not be determined; therefore, the extent of aerodynamic penalties resulting from this {{ce accumulation could not be determined. Also, determination of the aerodynamic effect of ice on 4 specific airfoil is not a requirement in the certification of the aircraft. Even if this were a requirement, it would be virtually impossible to determine the aerodynamic penalties of the various shapes and textures of ice that may be accumulated in flight.

[(a) FINDINGS] Conclusions (a) Findings 1, The aircraft was properly certificated and maintained in accordance with current requirements, with the exception of the unreported malfunctioning CVR, The crew was certificated and met the qualification criteria of current regulations, There was no preimpact failure cf the aircraft structure, systems, or powerplants, The aircraft weight and balance was well within that allowed and within 189 pounds of that listed on the completed weight and balance form, ‘leing conditions were accurately forecast, The flighterew should have reasonably expected to accumulate airfoil ice during the descent through the overcast approaching Sioux City, ‘The captain did not select airfoil anti-ice for the descent through the overcast. The requirement for the use of airfoil anti- ice under these conditions was clearly outlined I in the company operating manual, The aircraft landed at Stoux City with an accumulation of airfoil ice, The captain failed to have the aircraft deiced while at the ramp at Sioux City, although ground personnel advised him that ice was adhering to the aircraft, The captain. failed to recognize the aerodynanic penalties of airfoil icing, He did not persorally check, or require his first officer to personally check, the ice accumulation on the aircraft, although - he was advised of its presence, The captain did not select the proper takeoff thrust required for the existing gross weight of the aircraft. 13. A stall occurred near the top of ground effect as a result of aerodynamic and weight penalties of airfoil ice accumulation, - . wt ee an re “we Oe nei ae eam ON Ne Ne Am REN 14, The integrity of the cabin section, which was maintained throughout the impact path, prevented incapacitating injuries to the passengers, 15.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] The captain explained that as the ice did not cause hia any problem on approach and landing, he did not expect it to give him a probles on takeoff. _ He further stated that during his previous flying career in Douglas DC-3, Martin 404, and Fairchild F-27 atreraft, he had at tines mad: takeoffs with ice adhering to the airfoils, following an approach and landing with the ice, in which no difficulty was experienced, The Board further concludes that under the circumstancea, the aircraft should have been detced before takeoff, Studies, actual experience, training, and the regulations have long emphasized the aerodynamic penalties of takeoff with ice adhering to the airfoils, « 26 - NASA studies have determined that the magnitude of the aerodynamic penalties resulting from an iced airfoil were primarily a function of the shape and size of the ice formation near the leading edge of the airfoil, These studies also showed that rotation of an iced airfoil to angles of attack, other than that at which icing occurred, caused sufficiently large changes in the pitchitng-moment coefficient that, in flight, rapid corrections in trim might be required in order to avoid a hazardous situation. This would seem to be confirmed by the two training incidents in which unexpected violent rolls and delayed recovery were experfenced during approach to stalls with known or suspected airfoil ice, Although the takeoff gross weight of the aircraft required the use of -l power (14,000 pounds of thrust per engine), and this power requirement was clearly identified on the completed wefght and balance form given to the crew by the station agent,the crew selected <5 power (12,250 pounds of thrust per engine) for this takeoff, No explanation was offered by the crew for this power selection other than that they failed to recognize the gross weight power requirement.

[ANALYSIS AND CONCLUSIONS > ANALYSIS] No elevator trim was applied after 175 knots, The stick shaker stall warning was recei{{ved at 137 knots and a normal recovery was made using 20° of flaps, The next maneuver was an approach to stall in the 20° flap configuration while executing a 20° banked turn to the left, V was computed at 130 knots, no elevator trim was applied after 181 knots. The stick shaker stall warning was received at 114 knote, at which time the pilot applied takeoff power and rudder and aileron to roll out of the turn. The moment rudder was applied the right » 25 « The ceiling on arrival was 700 feet above the ground, According to the flight recorder, approximately Ll minutes were required for the descent through the overcast, Although mojerate to locally heavy mixed icing was forecast in clouds and precipitation, the captain elected not to use the airfoil anti-ice during penetration of this overcast; however, he did use engine anti-ice, The captain offered no explanation as to why he etected not to use airfoil antisice other than he did not think it was neeced, The Board finds this reasoning difficult to accept in that company policy, sound operating procedures, and good pilot judgment, in consideration of the existing weather conditions, dictated the selection of both engine and airfoil anti-ice, _ As an iced airfotl no longer retains the aerodynamic characteristics of the clean airfoil, the precise characteristics of the iced airfoil are somewhat unpredictable, NASA studies have confirmed that aerodynamic penalties do result in higher stall speeds and lower stall angles of attack, Therefore, the approach and landing at Sioux City, flown at the performance figures for a clean airfoil, were exposed to possible control difficulties, The approach and landing at Sioux City were most probably completed without incident because they were flown at near the same angle of attack as the angle of attack at which the ice was accunulated and the benefit of the increased lift as the aircraft descended into ground effect during the landing flare; whereas, during lift-off the aircraft was rotated to an angle of attack most probably 7° to 9° greater than the angle of attack at which the ice was accuaulated, coupled with the reduced lift as the aircraft was departing ground effect, Upon arrival at the ramp, and being advised that ice was adhering to the afreraft, the captain declined the station agents’ offer to defce it, Also, neither the captain nor the first officer left their seats to personally examine the reported ice.

[ANALYSIS] However, the airplane did fly through moderate rime icing during its descent for landing at CLE. Because the CVR began recording after the airplane landed at CLE, the Safety 8oard cannot determine if the flightcrew discussed the use of anti-ice protection during the descent. However, the flightcrew had received ample weather information, including a PIREP, about icing conditions around CLE during the approach, so they should have selected both wing and engine icing protection during the descent for landing. In fact, there is reference on the CVR to use of these systems while the airplane was on the ground at CLE. The Safety Board concludes that the flightcrew most likely selected the system "on." Further, there was no evidence to suggest that the anti-ice system was inoperative. The Safety Board considered the possibility that the pilots did use the wing anti-ice system during the descent and that moisture may have run aft of the heated wing surface and refrozen on the upper wing surface. However, information from the manufacturer indicates that the high temperatures (350° F or 1779 C) of the heated wing surface would vaporize any liquid moisture on the wing and that the “runback" was unlikely. Nevertheless, this possibility cannot be ruled out. The Safety Board believes that the most likely possibility of explaining the formation of ice on the wing surface is that the flightcrew used the wing anti-ice system during the approach and that the falling dry snow melted and refroze while the airplane was on the ground at CLE. The Safety Board believes that this scenario is possible because the wing would be “hot” upon touchdown (when the air/ground relay deactivates the anti-ice system automatically) and the blowing dry snow can melt on the wing and refreeze, as the wing temperature cools to belouw freezing. 2.3 Effect of Ice or Snow Contamination on Airplane Performance The airplane’s performance profile, which was developed during the investigation, and witness observations indicate that the takeoff roll and acceleration were normal wntil the airplane was rotated and lifted off the ground. The data showed that liftoff occurred at a slightly higher-thannormal airspeed and that the airplane began to climb. However, when reaching an altitude of 100 feet or less, the airplane rolled steeply.

[FINDINGS] If Ryan had fulfilted this obligation it would have become aware of the previous accidents involving wing ice contamination. Then Ryan would have been able to provide the training and guidance to its flightcrews that should have prevented this accident. Thus, the airline is also cited as a causal factor in the accident. 50 3.0 CONCLUSIONS 3.1 Findings J}}. There was no evidence of preexisting airplane structural, systems, or engine faults that contributed to the loss of contro) of the airplane 7 seconds after liftoff from runway 23L at Cleveland-Hopkins International Airport. 2. Four previous accidents of DC-9 series 10 airplanes, also invelving loss of control almost immediately after liftoff, were attributed to a loss of aerodynamic efficiency due to ice accumulation on the wings. 3. The accident airplane had flown through conditions conducive to the accumulation of moderate rime ice during the descent for landing at Cleveland-Hopkins International Airport, about 40 minutes before the accident, and the flightcrew probably used the wing anti-ice system during the descent. 4. Ground conditions at the Cleveland-Hopkins International Airport during the 35-minute turnaround were not conducive to airframe icing because of the dry snow and the low ambient temperatures; however, the melting and refreezing of snow on the previously heated wings could have produced = an accumulation of ice on the wing upper surface. The flightcrew did not exit the airplane to conduct an exterior preflight inspection at the Cleveland-Hopkins International Airport to verify that the wings were free of ice contamination, and a requirement for such an inspection was not specified in the Ryan DOC-9 Operations Manual. The first officer was controlling the airplane, and the takeoff rol] and rotation were normal and accomplished in accordance with prescribed procedures. 7. The liftoff occurred at a _ higher-than-normal airspeed; however, the lift-producing efficiency of the wing was degraded by contamination, and the stall speed margin at liftoff was minimal. There was some physical evidence but no evidence derived from the performance analysis to corroborate a tail strike at takeoff. However, a tail strike could occur with normal pilot procedures during an attempted takeoff with wing contamination. 51 The airplane’s wings stalled abruptly and without warning as the airplane began to climb and the aerodynamic advantage of ground effect diminished. At the time of stall, the airplane had sufficient speed to achieve a 1.4 g load factor with normal aerodynamic characteristics. The engine compressor Surges were caused by the disturbed airflow aft of the wince and at the engine inlet as the airplane approached stall. The steep roll concurrent with stall] was caused by the irregular lift distribution across the wing and was not controllable by the pilot, thereby preventing recovery. The DC-9 series 10 has no Wing leading edge lift augment ing devices and is particularly vulnerable to degraded aerodynamic performance as 2 result of minute amounts of wing contamination than the later model DC-9 and MO-80 airplanes that have leading edge devices.

[ANALYSIS] If in doubt, DE-ICE! The Safety Board concludes, from the observations of witnesses, that neither of the flightcrew menbers exited the airplane to conduct a walk around inspection or a close observation of the wing surface. further, the Safety Board concludes that the flightcrew did not violate written Ryan policy on this subject. The flightcrew may have observed the wing leading edge from the cargo loading door or the cockpit windows. However, in the existing Jighting condition and from that distance, the detection of a critical, but minute, amount of ice would have been unlikely if not impossible. The flightcrew may have been influenced by several factors in their decision to remain in the airplane. First, they may have believed that the air *.1as too cold to contain liquid water that could freeze and stick to the wing surface--the ambient temperature was 23°F and the 14-knot wind was blowing dry snow off of other objects visible to the crew. Little information is available regarding the possibility of ice forming during the melting/cooldown period following the deactivation of wing anti-ice systems after landing. Second, the Safety Board noted that both crewmembers’ experience prior to flying with Ryan was in OC-9 series 30 aircraft. The captain flew the C-9 in the U. S. Air Force, and the first officer flew the DC-9-30 and DC-9-50 at USAir. Because these models have leading edge devices, they are not as vu,oerable as the OC-9 series 10 airplane to critical performance degradation “rom small amounts of wing contamination. Therefore, even if the captain or first officer had encountered similar weather conditions prior to flying with Ryan, they most likely would not have encountered contro! problems and their concerns about the hazard of wing ice contamination would probably have been ftessened. Another consideration bearing on the crew’s attention to the possibility of ice is the lack of de-icing activity by other operators. De-icing equipment had been standing by for approximate'y 1-1/2 hours and was immediately available. There was no evidence of fiscal or schedule pressures by the airline that would have discouraged the crew from using that equipment. The Safety Board also considered the possibility that fatigue influenced the pilots’ judgment during the ground operations at CLE and their decision not to conduct an exterior preflight inspection of the 47 airplane. The flightcrew had flown the same nighttime schedule for 6 days, including the night of the accident, between BUF and IND with an intermediate stop each way in CLE. The captain had flown six successive night flights on the same 8UF-CLE-IND and return route the week before the accident. He had 1 day off between the two periods of duty. The six flights, averaging about 3.8 hours each night, did not exceed FAA maximum flight time limitations; however, the captain’s schedule hai recently increased from the routine of flying for 5 days, followed by 9 days off-duty time at home in California. Although his family said that he was accustomed to night flying, the recent increase in duty and flight time could have induced fatigue.

[ANALYSIS] The Safety Board therefore concludes that efforts to educate line pilots of OC-9 series 10 airplanes about this problem have not been adequate. There are many reasons for the inadequacy of these efforts. Much of the written material has been presented to airline management. There has been general agreement on the accuracy of the data, but no real understanding of the significance of the problem has been evident. Even in cases where the significance is understood, line pilots are apparently not giving the problem the attention that it merits. 49 Accumulations of ice as thin as 0.015 inch on the wings of a DC-9 can reduce the stall angle of attack below stall warning activation. Investigators have ‘ound that the vast majority of OC-9 series 19 pilots questioned are either unaware of these facts or jack an appreciation for the criticality of visually imperceptible amounts of wing contamination. The Safety Board is concerned that when aircraft are sold, or when there are changes of pilots and instructors, an opportunity exists tor the loss of “corporate memory" of the significance of the icing problem on the DC-9. Although Douglas has issued material and urged that the wing icing problem be incorporated into the airplane flight manuals, they took no positive action to do so. By including the information in the approved Airplane Flight Manual, it would be directly available to the line pilots, and it would be transferred with the ownership of an aircraft when it is sold to a new operator. Ryan acquired eight DC-9s in 1989 and was unaware of the critical icing information until after the accident. If the information had been contained in the approved Airplane Flight Manual, the subject would have been emphasized in Ryan’s initial training of its pilots. Thus, the Safety Board believes that after four previous accidents, sufficient knowledge has existed within both the FAA and Douglas on the high vulnerability of the DC-9 series 10 to flight control problems in freezing weather conditions and that this information should have been disseminated in such a manner that it would be available to all of the pilots of these airplanes. The FAA could have required, and Douglas could have prov‘ded, additional information about this problem in the approved Airplane Flight Manual. Their failure to do so is a causal factor in this accident. Similarly, the Safety Board believes that any operator acquiring a new model airplane in its fleet has an obligation to request from the manufacturer, and any other available sources, information unique to the safe operation of that airplane. If Ryan had fulfilted this obligation it would have become aware of the previous accidents involving wing ice contamination. Then Ryan would have been able to provide the training and guidance to its flightcrews that should have prevented this accident. Thus, the airline is also cited as a causal factor in the accident. 50 3.0 CONCLUSIONS

[FINDINGS] The steep roll concurrent with stall] was caused by the irregular lift distribution across the wing and was not controllable by the pilot, thereby preventing recovery. The DC-9 series 10 has no Wing leading edge lift augment ing devices and is particularly vulnerable to degraded aerodynamic performance as 2 result of minute amounts of wing contamination than the later model DC-9 and MO-80 airplanes that have leading edge devices. The flightcrew had not been given specific training or other educational materia! to inform them of the more critical effects of wing contamination on DC-9 series 10 airplanes. The Doug’as Aircraft Company has issued numerous articles on the subject of wing contamination, but there is no system to ensure that the critical information reaches all line pilots of these airplanes. Both the FAA and Douglas Aircraft Company have been aware for several yeas of the propensity of the OC-9 series 10 to the loss of control caused by wing contamination, but neither of them took positive action to include related information in the approved Airplane Flight Manual. Had additional information or cautions about the high vulnerabil:ty of the DC-9 series 10 to loss of control caused by wing contamination been placed in the approved Airplane Flight Manual, it would have been available to line pilots. Ryan International Airlines had the Opportunity and obligation to request information relating to previously identified Safety issues when it acquired the OC-9 airplanes in 1989 but failed to do so. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the failure of the flightcrew to detect and remove ice contamination on the airplane’s wings, which was largely a result of a lack of appropriate response by the Federal Aviation Administration, Douglas Aircraft Company, and Ryan International Airlines to 52 the known critical effect that a minute amount of contamination has on the Sta'l characteristics of the DC-9 series 10 airplane. The ice contamination led to wing stall and loss of control during the attempted takeoff.

[CONCLUSIONS] CONCLUSIONS Findings RECOMMENDATIONS APPENDIXES Appendix A--Investigation and Hearing Appendix B--Cockpit Voice Recorder Transcript Appendix C--Personnel Information Appendix D--Additional Aircraft Information Appendix E--Douglas Aircraft Company Letter Report Appendix F--Douglas Aircraft Company Graphic J}}lustration . Appendix G--Company History Provided by Ryan International Airlines, Inc. EXECUTIVE SUMMARY About 0019, Sunday, February 17, 1991, Ryan International Airlines flight 590 (Ryan 590), a OC-9 series 10 airplane, crashed while taking off from Cleveland-Hopkins International Airport. The flightcrew consisted of two pilots. There were no other crewmembers or passengers on the flight, which was contracted to carry mail for the U.S. Postal Service. Both pilots were fatally injured, and the airplane was destroyed as a result of the accident. The Nationat Transportation Safety Board determines that the probable cause of this accident was the failure of the flightcrew to detect and remove ice contamination on the airplane’s wings, which was largely a result of a lack of appropriate response by the Federal Aviation Administration, Douglas Aircraft Company, and Ryan International Airlines to the known critical effect that a minute amount af contamination has on the stall characteristics of the DOC-9 series 10 airplane. The ice contamination led to wing stall and loss of control during the attempted takeoff. The safety issues discussed in this report include’ the dissemination of information regarding precautions to be taken when operating in conditions conducive to airframe ice and the particular susceptibility of DC-9 series 10 airplanes to control problems during take off when a minute amount of ice is on the wing. z t s £ i ¥ rf NATIONAL TRANSPORTATION SAFETY BOARD WASHINGTON, D.C. 20594 AIRCRAFT ACCIDENT REPORT RYAN INTERNATIONAL AIRLING DC-9-15, N565PC LOSS OF CONTROL ON TAKEOFF CLEVELAND-HOPKINS INTERNATIONAL AIRPORT, OHIO FEBRUARY 17, 1991 1. FACTUAL INFORMATION 1.1 History of the Flight About 0019, Sunday, February 17, 1991, Ryan International Airlines flioht 590 (Ryan 590), a ODC-9-15, crashed while taking off from Cleveland-Hopkins International Airport (CLE). The flightcrew consisted of two pilots. There were no other crewmembers or passengers on the flight, which was contracted to carry mail for the U.S. Postal Service. Both pilots were fatally injured, and the airplane was destreyed as a result of the accident.

[ANALYSIS] At the Safety Board’s public hearing on this accident, a representative from McDonnell-Douglas stated that small amounts of upper wing ice may severely degrade the lifting capability of the wing and lead to loss of roll and pitch control on DC-9-10 series airplanes. He concluded that the DC-9-10 series and other airplanes, with and without leading edge slats, would be affected to varying degrees by small amounts of upper wing ice contamination. For example, granular ice of only 0.030 inch (similar to the roughness of 30-40 grit sandpaper) would degrade. the maximum lifting capability of the DC-9 wing by about 20 percent. For a given increase in angle of attack, an ice contaminated wing would have a lesser increase of lift than would an ice-free wing. The stall speed would increase and the stall angle of attack would decrease, possibly to the point that the stall warning indicator (receiving its signals from angle of attack sensors, not airspeed sensors) would not activate before stall. Indeed, in the case of flight 1713, no stick shaker was heard on the CVR tape, although the airplane was in the stall regime before impact. In addition, if less than normal lift is available during the takeoff pitch rotation, the airplane may not be able-to leave the ground either when expected or in a stable manner. In any case, the stall safety margin is significantly reduced. , Ice contamination also may produce roll oscillations and unexpected pitch-up tendencies during flight. Ice accumulations usually are not uniform and result in nonuniform lift degradations on the wings, horizontal tail, and, to a small degree, the fuselage. For example, a small section of ice on an otherwise contaminant-free wing or a small section of rougher ice on a contaminated wing, may be the first area on the wing to stall or produce less than normal lift. This uneven lift may result in the onset of roll, followed by pilot initiated counter ailreon and spoiler deflections which can quickly set up roll oscillations. On swept wing airplanes, contaminated outboard wing areas also can produce unexpected pitch-up tendencies because the outboard wing areas are usually behind the center of gravity of the airplane. When the wingtips stall; the inboard parts of the wings (ahead of . the center of gravity) produce proportionally more lift and the nose pitches up. However, the 35 greater than normal pitch rate on flight 1713 was present during initial rotation (when the wings were unloaded) indicating that the high pitch rate was pilot-induced. Ice-induced pitch rates, on the other hand, result from loaded wings that just reach the localized stall angle of attack. The Safety Board is not aware of any service history or pilot reports describing DC-9-10 series ice-induced pitchup tendencies. The small amount of ice on the wings of the airplane contributed to significant controllability problems on flight 1713. Safety Board calculations show that a stall could have occurred on the accident airplane at 165 knots calibrated airspeed with 1.4 Gs on the airframe if there: had been about a 20 percent reduction in maximum lifting capability.

[PROBABLE CAUSE] By applying a maximum effective strength glycol solution after deicing, anti-ice protection could have been increased by a time factor of 2.8 over the 38 percent glycol solution used on flight 1713. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the captain's failure to have the airplane deiced a second time after a delay before takeoff that {{ed to upper wing surface contamination and a loss of.control during. rapid takeoff rotation by the first officer. Contributing to the accident were the absence of regulatory or management controls governing operations by newly qualified flight crewmembers and the confusion that existed between the flightcrew and air traffic controllers that led to the delay in departure. - - 45 ’ 4. RECOMMENDATIONS Asa result of its investigation, the Safety Board made the following recommendations: ~-to the Federal Aviation Administration: Until such time that guidelines for detecting upper wing surface icing can be incorporated into the airplane flight manual, issue an air carrier operations bulletin directing all principal operations inspectors to require that all McDonnell Douglas DC-9-10 series operators anti-ice airplanes with maximum effective strength glycol solution when icing conditions exist. (Class II, Priority Action) (A-88- 134) - Expedite the evaluation of the effectiveness of Association of European Airlines guidelines concerning the use of European types I and II deicing and anti-icing fluids. If European methodology is more effective than current U.S. methodology, incorporate their guidelines into the next version of Advisory Circular 200-17. (Class Il, Priority Action) (A-88- 135) Require all DC-9-10 series operators to establish detailed procedures for detecting upper wing ice before takeoff. (Class !!, Priority Action) (A-88- 136) Establish minimum experience levels for each pilot-in-command and second-incommand pilot, and require the use of such criteria to prohibit the pairing on the same flight of pilots who have less than the minimum experience in their respective positions. (Class I!, Priority Action) (A-88- 137) Amend 14 CFR 121.434 to require that a second-in-command pilot complete initial operating experience for that position while actually performing the duties of a second-in-command under the supervision of a check pilot. (Class II, Priority Action) (A-88- 138) Review and revise, as necessary, the engineer performance standards ‘for appropriate airports to account for the reduced airport capacities that occur when deicing operations are in progress. (Class Il, Priority Action) (A-88- 139) Initiate a research project to acquire data from dedicated sensors to determine what consideration, if any, should be given to wake vortices in a parallel offset runway situation. (Class Il, Priority Action) (A-88-140) Require commercial! operators to conduct substantive background checks of pilot applicants which include verification of personal flight records and examination of training, performance, and disciplinary records of previous employers and Federal Aviation Administration safety and enforcement records. (Class Il, Priority Action) (A-88-141) Issue an airworthiness directive to require more complete operating instructions on the exterior side of the tailcone exit hatch of DC-9 airplanes.

[ANALYSIS] Surges also have been noted on CVR recordings from accidents with no wing contamination. These surges occurred when the intakes: were no longer aligned with the relative wind during dynamic maneuvering of the airplane, causing compressor blades to stall and subsequent surges. In all instances, the surges were the effect rather than the cause. Consequently, engine compressor surges were not a causal factor in this accident. The lack of leading edge devices on the wings of the DC-9 airplane make it more vulnerable to performance degradation due to wing contamination; however, the Safety Board believes that the FAA and McDonnell-Douglas have adequately warned DC-9-10 series operators of such possible degraded flight characteristics through AC 20-117 and several articles on airframe contamination in McDonnell-Douglas publications.7 In general, McDonnel!-Douglas provided guidance for carefully inspecting for “almost undetectable amounts of ice," and the FAA regulations require that airfoil 7Brumby, Wing Surface Roughness--Cause and Effect, and Brumby, Ralph E. Aerodynamics.and Cold Weather Operations in DC Flight Approach #41, Flight Test and Operations Group, Douglas Aircraft Company, McDonnell-Douglas Corporation, Long Beach, California, December 1982. 36 surfaces be free of contamination” before takeoff, which is adequately specific information for operators. Finally, airplane certification requirements for performance are based on airfoil surfaces that are not contaminated by ice, snow, or frost. The Safety Board has investigated three previous DC-9-10 series icing-related accidents which were similar-to the circumstances of the accident involving flight 1713.8 In two of the accidents, ice was visible to the crews before takeoff; in the other accident, the crew failed to examine the wings before takeoff. The Safety Board believes that the November 15, 1987, accident again demonstrates that even small amounts of contamination on the upper surfaces of an airplane can seriously degrade lift. This accident underscores the critical importance for the pilot-in-command to ensure the surfaces are clean before every takeoff when in conditions conducive to contamination. The crew of flight 1713 also failed to examine the wings for contamination before takeoff. Therefore, the Safety Board believes that there is no justification for questioning the FAA certification of a DC-9-10 series airplane. 2.8 The First Officer's Actions During Rotation The first officer's poor rotation technique probably contributed to the loss of airplane control. Evidence of trouble during the takeoff rotation was apparent from data recovered from the FDR. The altitude dip associated with pitch rotation in a DC-9-14 airpiane is normally about 50 to 60 feet below field elevation, consistent with a pitch angle of about 6° during liftoff. Under normal . circumstances, the magnitude of the dip is porportional to the pitch attitude of the airplane while it is still on the ground. The pitch rate defines the initial slope of the dip. For the accident flight, the dip was about 120 feet, indicating a pitch attitude of about 14° while the airplane was very close to the ground.

[ANALYSIS] The accumulated precipitation on the upper horizontal surfaces of the airplane probably would have been a combination of snow and melting snow or slush. Consequently, because of the dilution of the anti-icing fluid, the actual thickness of the slush probably would have been slightly greater than the water equivalent of the snow alone and would have frozen into a roughened surface. Even this modest amount of surface roughness on the wings of a DC-9-10 series wing could cause controllability problems according to McDonnel!-Douglas. 34 The contamination of the airframe surfaces of flight 1713, as thin as it may have been, could have been delayed if the airplane had been anti-iced following the deicing. According to the Association of European Airlines, a full-strength glycol anti-icing application would have prevented any ice buildup 2.8 times longer than the 38 percent glycol deicing a application that flight 1713 received. Federal guidelines concerning deicing fluid type, temperature, consistency, and application methods are summed up in FAA AC 20-117. The AC thoroughly discusses deicing methodology in general use in the United States. It does not, however; incorporate more advanced deicing and antiicing methods using “type II” deicing fluids that have been used by European countries for several years. The 1986 edition of the Association of European Airlines.Recommendations for De-/Anti-Icing of Aircraft on the Ground includes specifications for ground deicing fluids, fluid dispensing equipment, quality control guidelines and procedures,. application procedures and methods of ensuring proper interaction and communication between maintenance and flightcrews. The Safety Board acknowledges that the FAA, in conjunction with the Air Transport Association (ATA) and the Society for Automotive Engineers (SAE), is actively studying the advantages and disadvantages of the use of type II deicing fluids. Also, the Board notes that several U.S. manufacturers are now experimenting with other forms of advanced deicing and anti-icing systems and new mechanical ice detecting devices for aircraft. The Board encourages expedited research and testing in this area, under the sponsorship of the FAA. Also, the Board believes that, should type Ii or other advanced fluids prove safe for U.S. operations, their use should be highly encouraged by the FAA. 2.6 Aerodynamic Effects of Airframe Contamination and the Results of the Contamination on Flight 1713 The Safety Board believes that ice contamination that formed on flight 1713 during the 27 minutes it waited to depart Stapleton was sufficient to raise the stall speed of the airplane and compromise its stability and the pilot's ability to maintain control. At the Safety Board’s public hearing on this accident, a representative from McDonnell-Douglas stated that small amounts of upper wing ice may severely degrade the lifting capability of the wing and lead to loss of roll and pitch control on DC-9-10 series airplanes. He concluded that the DC-9-10 series and other airplanes, with and without leading edge slats, would be affected to varying degrees by small amounts of upper wing ice contamination.

[ANALYSIS] It should be noted that little wake vortex data exists forthe B-767 and there is some possibility that its wake vortex may be longer lasting than its weight would suggest, although not as long lasting or as strong as the Boeing 747. Therefore, it is conceivable, but unlikely, that the B-767 could have produced a wake vortex.strong enough to affect a DC-9 to some unknown degree, from a lateral distance of 1,600 feet. This could have occured if, and only if, all the other conditions previously cited for potential encounter were present. , However, the Safety Board wishes to emphasize that it has not eliminated the possibility that on a different day with different conditions and different aircraft, a potential problem might exist concerning wingtip vortices: Therefore, the Board believes that the FAA should commence a research project to acquire data from dedicated sensors to determine what consideration, if any, should be given to wake vortices in a parallel offset runway situation. 2.5 Airplane Deicing and Subsequent Contamination The Safety Board believes that the airplane was adequately deiced before it departed the deice pad. Evidence suggests that the combination system of fixed deicing snorkels and mobile deicing trucks used by Continental at Denver is quicker and more efficient than the use of deicing trucks alone. Nevertheless, since the airplane was exposed to a moderate snowstorm in subfreezing conditions for approximately 27 minutes following deicing, the Safety Board believes that portions of the airframe became contaminated with a thin, rough layer of ice. The pilot of Continental flight 875 stated that he did not see any contamination on the wings of flight 1713. However, several surviving passengers on flight 1713 reported seeing some “ice” on engine inlets or in “patches” on the wing after deicing. These accounts suggest isolated fragments of contamination. During precipitation in subfreezing ambient temperatures, ice can accumulate on airframe surfaces after a thorough deicing when the deicing solution evaporates, runs off, or is diluted with the precipitation. All three of these conditions occurred on the wings of flight 1713, with dilution of I the deicing solution having been the predominant condition. Due to many variables involved, the Safety Board found it impossible to determine exactly where or exactly how much ice had formed on the wing and empennage surfaces of flight 1713. The Safety Board believes that enough wet snow (0.29 inch) fell on flight 1713 during the 27 minutes between deicing and takeoff to dilute the deicing fluid to the point where ice began to reform. This 0.29 inch of snow, if melted, would equate to about 0.032 inch of water. The accumulated precipitation on the upper horizontal surfaces of the airplane probably would have been a combination of snow and melting snow or slush. Consequently, because of the dilution of the anti-icing fluid, the actual thickness of the slush probably would have been slightly greater than the water equivalent of the snow alone and would have frozen into a roughened surface.

[ANALYSIS] The small amount of ice on the wings of the airplane contributed to significant controllability problems on flight 1713. Safety Board calculations show that a stall could have occurred on the accident airplane at 165 knots calibrated airspeed with 1.4 Gs on the airframe if there: had been about a 20 percent reduction in maximum lifting capability. Flight 1713's maximum airspeed of about 165 knots was recorded on the FDR simultaneously with 1.4 Gs. At almost exactly the same time, an exclamation from a crewmember was recorded on the CVR. A 20 percent reduction in lift would have resulted from 0.03 inch of ice, which the Safety Board believes is at least the amount that could have accumulated in 27 minutes. Therefore, the Safety Board concludes that the accident was precipitated by the captain's failure to return for a second deicing after the extensive delay before takeoff because the upper wing surface contamination: that existed was sufficient to cause the loss of control during the takeoff attempt. 2.7 Airplane Maintenance and Certification The airplane was maintained in accordance with current Federal regulations and Continental maintenance policies. All airworthiness directives and service bulletins had been complied with. The one significant open discrepancy present at the time of the accident (the inoperative center fuel tank quantity guage) did not contribute to the accident in any way because the center fuel tank contained only residual fuel during the flight. The installation of red “STALL” lights on the glare shield of the airplane instead of amber “STALL WARN” lights was not in accordance with the original engineering order. In addition, the lack of a “Stall Comparator Failure” light on the annunciator panel was contrary to the engineering order. The filaments of both red “STALL” lights were found stretched, indicating that they were illuminated at impact. The Safety Board believes that the fact that the lights were red instead of amber and labeled incorrectly «was not a causal factor in the crash. Also, the lack of a “Stall Comparator Failure Light” was not a causal factor in the crash. The engine compressor surges noted on the CVR tape during the last seconds of the flight before impact were attributed to aerodynamic factors and not mechanical failures. Such surges have been noted in past accidents and incidents where the upper wing surfaces were contaminated and disturbed airflow from the wings entered the engine intake. Surges also have been noted on CVR recordings from accidents with no wing contamination. These surges occurred when the intakes: were no longer aligned with the relative wind during dynamic maneuvering of the airplane, causing compressor blades to stall and subsequent surges. In all instances, the surges were the effect rather than the cause. Consequently, engine compressor surges were not a causal factor in this accident.

[ANALYSIS] The flightcrew's apparent lack of concern regarding the prolonged operations in icing conditions may have been influenced by their extensive experience of safely flying commuter aircraft in winter weather conditions, especially in icing conditions that are prevalent in the Great Lakes region. In addiction, they were probably confident in the ability of the airplane deicing system to adequately shed the ice that had been accumulating on the wings and in their ability to perform safely under the existing circumstances. The flightcrew was operating in icing conditions that exceeded the limits set forth in 14 CFR Part 25, Appendix C, resulting in a complete loss of aircraft control. However, the insidious nature of these icing conditions was such that the ice accumulation on the observable portions of the wings, windshield and other airframe parts was most likely perceived by the flightcrew as nonthreatening throughout the holding period. Moreover, the flightcrew was undoubtedly unaware that the icing conditions exceeded the Appendix C limits and most likely had operated in similar conditions many times prior to the accident, since such conditions occur frequently in the winter throughout the Great Lakes and northeastern parts of the United States. 193 Further, the flightcrew entered the holding pattern with the belief that the holding would be of a short duration, unaware that it would be continually extended in short increments for a total of 39 minutes. Therefore, the Safety Board concludes that if a significant amount of ice had accumulated on the wing leading edges so as to burden the ice protection system, or if the crew had been able to observe the ridge of ice building behind the deice boots or otherwise been provided a means of determining that an unsafe condition could result from holding in those icing conditions, it is probable that they would have exited the conditions. The Safety Board is aware that ATR provided information to the operators of its airplanes that indicated that encounters with certain freezing precipitation conditions could result in “roll axis anomalies.” However, this information was vague and did not indicate to flightcrews how to determine that they were in freezing rain, nor did it specifically alert them that encounters with freezing rain could result in sudden autopilot disconnects, uncommanded aileron movements and rapid roll excursions. Therefore, the crew of flight 4184 had no reason to expect that the icing conditions in which they were operating could cause the autopilot to disconnect unexpectedly because of the onset of an aileron hinge moment reversal and cause a loss of normal, stable aileron control. In addition, ATR did not provide, nor did the regulatory authorities require, training or guidance to pilots about the roll axis “anomalies” or the recovery techniques if such an event should occur. Thus, the Safety Board concludes that the crew of flight 4184 was not provided with adequate information by the manufacturer or the regulatory authorities to recognize and cope with the problems they experienced during an encounter with freezing rain.

[ANALYSIS] Because there was no discussion between the crewmembers regarding the 1533:56 caution alert, it is possible that it was activated by one of the other aircraft systems. Assuming the chime at 1533:56 was activated by the ice detection system, consistent with the drag increase noted at that time on the FDR, the Safety Board concludes that the flightcrew’s failure to increase the propeller RPM to 86 percent and activate the Level III ice protection system was not a factor in the accident because: 1) according to ATR, the increased propeller RPM is necessary to increase the ice shedding capabilities of the propeller blades; 2) the increased propeller RPM will prevent the formation of ice aft of the deice boots in the area of the propeller slipstream, but will not prevent the formation of ice on the wing in the areas behind the deice boots, in front of the ailerons, or the airflow over the ailerons. There was no discussion recorded on the CVR to suggest that the flight crewmembers had a safety concern about the icing conditions in which they were 191 holding. Two comments by the crewmembers recorded on the CVR indicated they were aware that ice was accreting on the airframe. The first comment, “I’m showing some ice,” occurred about 9 minutes before the initial upset of the airplane, and the second comment, “we still got ice” occurred about 2 minutes before the upset. Neither comment indicated the type or amount of ice, nor did the comments suggest that the crew was aware the ice accretions were related to an encounter with freezing drizzle or freezing rain. The comments only indicate that the flightcrew was aware that they were operating in an icing environment. Further, the flightcrew responded appropriately to the caution alert at 1541:07 by increasing the propeller RPM to 86 percent and activating the deice boots 5 seconds later, or about 16 minutes before the upset. Although AMR Eagle cautioned pilots in its 1989 memorandum about flight in freezing rain, the information and training provided by Simmons Airlines/AMR Eagle to its flightcrews did not prohibit holding in icing conditions that were perceived to be within the capabilities of the airplane. In addition, the crew did not have specific training or information necessary to determine that the airplane was operating in conditions (freezing rain) beyond those for which it had been certificated. It is generally understood that flight in icing conditions in any aircraft at low airspeeds with the flaps and/or the landing gear extended for a long period of time is not a good operating practice because such exposure increases the likelihood of a significant accumulation of ice on the flaps and/or landing gear, which could result in increases in weight and drag, and a decrease in aircraft performance.

[PROBABLE CAUSE] Although both crew members of flight 4184 were certified flight instructors, this was probably the first time they had experienced such unexpected and excessive roll and pitch attitudes in the ATR 72. If the operators had been required to conduct unusual attitude training, the knowledge from this training might have assisted the flightcrew in its recovery efforts and might have prompted the captain to provide useful information to the first officer to facilitate a timely recovery of the airplane. 210 3.2 Probable Cause The National Transportation Safety Board determines that the probable causes of this accident were the loss of control, attributed to a sudden and unexpected aileron hinge moment reversal that occurred after a ridge of ice accreted beyond the deice boots because: 1) ATR failed to completely disclose to operators, and incorporate in the ATR 72 airplane flight manual, flightcrew operating manual and flightcrew training programs, adequate information concerning previously known effects of freezing precipitation on the stability and control characteristics, autopilot and related operational procedures when the ATR 72 was operated in such conditions; 2) the French Directorate General for Civil Aviation’s (DGAC’s) inadequate oversight of the ATR 42 and 72, and its failure to take the necessary corrective action to ensure continued airworthiness in icing conditions; and 3) the DGAC’s failure to provide the FAA with timely airworthiness information developed from previous ATR incidents and accidents in icing conditions, as specified under the Bilateral Airworthiness Agreement and Annex 8 of the International Civil Aviation Organization. Contributing to the accident were: 1) the Federal Aviation Administration’s (FAA’s) failure to ensure that aircraft icing certification requirements, operational requirements for flight into icing conditions, and FAA published aircraft icing information adequately accounted for the hazards that can result from flight in freezing rain and other icing conditions not specified in 14 Code of Federal Regulations (CFR) Part 25, Appendix C; and 2) the FAA’s inadequate oversight of the ATR 42 and 72 to ensure continued airworthiness in icing conditions. 211

[CONCLUSIONS] The airplane was in a holding pattern and was descending to a newly assigned altitude of 8,000 feet when the initial roll excursion occurred. The airplane was destroyed by impact forces; and the captain, first officer, 2 flight attendants and 64 passengers received fatal injuries. Flight 4184 was a regularly scheduled passenger flight being conducted under 14 Code of Federal Regulations, Part 121; and an instrument flight rules flight plan had been filed. The National Transportation Safety Board determines that the probable causes of this accident were the loss of control, attributed to a sudden and unexpected aileron hinge moment reversal that occurred after a ridge of ice accreted beyond the deice boots because: 1) ATR failed to completely disclose to operators, and incorporate in the ATR 72 airplane flight manual, flightcrew operating manual and flightcrew training programs, adequate information concerning previously known effects of freezing precipitation on the stability and control characteristics, autopilot and related operational procedures when the ATR 72 was operated in such conditions; 2) the French Directorate General for Civil Aviation’s inadequate oversight of the ATR 42 and 72, and its failure to take the necessary corrective action to ensure continued airworthiness in icing conditions; and 3) the French Directorate General for Civil Aviation's failure to provide the Federal Aviation Administration with timely airworthiness information developed from previous ATR incidents and accidents in icing conditions, as specified under the Bilateral Airworthiness Agreement and Annex 8 of the International Civil Aviation Organization. Contributing to the accident were: 1) the Federal Aviation Administration’s failure to ensure that aircraft icing certification requirements, operational requirements for flight into icing conditions, and Federal Aviation Administration published aircraft icing information adequately accounted for the hazards that can result from flight in freezing rain and other icing conditions not specified in 14 Code of Federal Regulations, Part 25, Appendix C; and 2) the Federal Aviation Administration's inadequate oversight of the ATR 42 and 72 to ensure continued airworthiness in icing conditions. viii The safety issues in this report focused on communicating hazardous weather information to flightcrews, Federal regulations regarding aircraft icing and icing certification requirements, the monitoring of aircraft airworthiness, and flightcrew training for unusual events/attitudes. Safety recommendations concerning these issues were addressed to the Federal Aviation Administration, the National Oceanic and Atmospheric Administration, and AMR Eagle. Also, as a result of this accident, on November 7, 1994, the Safety Board issued five safety recommendations to the Federal Aviation Administration regarding the flight characteristics and performance of ATR 42 and ATR 72 airplanes in icing conditions.

[CONCLUSIONS > FINDINGS] The 14 Code of Federal Regulations (CFR) Part 25, Appendix C, envelope is limited and does not include conditions of freezing drizzle or freezing rain; thus, the current process by which aircraft are certified using the Appendix C icing envelope is inadequate and does not require manufacturers to sufficiently demonstrate the airplane's capabilities in all the possible icing conditions that can, and do, occur in nature. 14. No airplane should be authorized or certified for flight into icing conditions more severe than those to which the airplane was subjected in certification testing, unless the manufacturer can otherwise demonstrate the safety of flight in such conditions. 15. If the FAA had acted more positively upon the Safety Board's aircraft icing recommendations issued in 1981, this accident may not have occurred. 205 16. ATR 42 and 72 ice-induced aileron hinge moment reversals, autopilot disconnects, and rapid, uncommanded rolls could occur if the airplanes are operated in near freezing temperatures and water droplet median volume diameter (MVDs) typical of freezing drizzle. 17. At the initiation of the aileron hinge moment reversal affecting flight 4184, the 60 pounds of force on the control wheel required to maintain a wings-level-attitude were within the standards set forth by the Federal Aviation Regulations. However, rapid, uncommanded rolls and the sudden onset of 60 pounds of control wheel force without any warning to the pilot, or training for such unusual events, would most likely preclude a flightcrew from making a timely recovery. 18. ATR is considering design changes to the lateral control system for current and future ATR airplanes that will reduce the susceptibility to flow separation-induced aileron hinge moment reversals. Such design changes could minimize the reliance on the changes to flight operations and pilot training that have already been mandated. 19. The French Directorate General for Civil Aviation (DGAC) and the Federal Aviation Administration (FAA) failed to require the manufacturer to provide documentation of known undesirable postSPS [stall protection system] flight characteristics, which contributed to their failure to identify and correct, or otherwise properly address, the abnormal aileron behavior early in the history of the ATR icing incidents. 20. The addition of a test procedure, similar to the "zero G" flight test maneuver (pushover) designed to identify ice-induced elevator hinge moment reversals, could determine the susceptibility of an aircraft to aileron hinge moment reversals in both the clean and iced-wing conditions and could help prevent accidents such as the one involving flight 4184. 21. Prior to the Roselawn accident, ATR recognized the reason for the aileron behavior in the previous incidents and determined that ice accumulation behind the deice boots, at an AOA sufficient to cause an airflow separation, would cause the ailerons to become unstable. Therefore, ATR had sufficient basis to modify the airplane and/or 206 provide operators and pilots with adequate, detailed information regarding this phenomenon. 22.

[ANALYSIS] It is not hazardous even though deicing/anti-icing equipment is not utilized unless encountered for an extended period of time (over 1 hour); 2) Light - The rate of accumulation may create a problem if flight is prolonged in this environment (over 1 hour). Occasional use of deice/anti-icing equipment removes/prevents accumulation. It does not present a problem if the deicing/anti-icing equipment is used; 3) Moderate - The rate of accumulation is such that even short encounters become potentially hazardous and use of deicing/anti-icing equipment or flight diversion is necessary; 4) Severe - The rate of accumulation is such that deicing/antiicing equipment fails to reduce or control the hazard. Immediate flight diversion is necessary. While these icing severity definitions provide some basis for assessing ice accumulation in PIREPs, they are subjective and are of limited use to pilots of different aircraft types. For example, using these definitions, "light" icing for a Boeing 727 could be "severe" icing for an ATR 72 or a Piper Malibu. The icing report provided by the captain of the A-320 Airbus that was holding at the HALIE intersection, near Roselawn, indicated that he observed about 1 inch of ice accumulate rapidly on his aircraft's icing probe. The captain provided a PIREP to ATC and reported the icing as "light rime." He stated in an interview after the accident that the anti-ice equipment on the airplane "handled the icing adequately," and he believed the icing intensity to have been "light to moderate." 166 The Safety Board concludes that icing reports based on the current icing severity definitions may often be misleading to pilots, especially to pilots of aircraft that may be more vulnerable to the effects of icing conditions than other aircraft. The Safety Board believes that the FAA should develop new aircraft icing intensity reporting criteria that are not subjective and are related to specific types of aircraft. In addition, the investigation revealed a problem with the aviation community's general understanding of the phrase "icing in precipitation," which is used by the NWS but is not defined in any aeronautical publications, including advisory circulars (ACs), Part 1 of the Federal Aviation Regulations or the Aeronautical Information Manual (AIM). This phrase is often contained in in-flight weather advisories; however, it does not typically specify types of precipitation. According to the NWS, this phrase is intended to include freezing drizzle and freezing rain. The Safety Board concludes that defining “icing in precipitation” in such publications would make pilots and dispatchers more aware of the types of precipitation and icing conditions that are implied by this phrase. Therefore, the Safety Board believes that the FAA should provide a definition of the phrase "icing in precipitation" in the appropriate aeronautical publications.

[ANALYSIS] This investigation has revealed that the ATR 42 and 72 were not required to be tested throughout a significant portion of the icing conditions that are specified in the Appendix C icing envelope. The limited number of test points accepted by the FAA as sufficiently comprehensive were well within the boundaries of the envelope and did not include the warmer, near freezing conditions at the upper boundary of the Appendix C envelope in which run-back icing and asymmetric sliding/shedding are likely to occur. Thus, by allowing limited data well within the envelope to suffice for certification purposes, the FAA effectively precluded any chance of identifying the phenomena that led to flight 4184's ice ridge buildup, uncommanded aileron deflection and loss of control. The Safety Board's concern about the adequacy of Appendix C criteria was heightened by the results of one December 1994 ATR icing tanker test in which ice accumulated behind the active portion of the ATR 72's deice boots during exposure to water droplet sizes of only 57 microns MVD, which is only slightly outside the Appendix C envelope. Further, data developed by NACA, the NASA predecessor, indicated in the 1950s that MVDs of 70 microns or more could be encountered in layer clouds. Flight in layer clouds is not an unusual event in this 171 country, but flight into layer clouds can result in encounters with icing conditions beyond those set forth in 14 CFR Part 25, Appendix C. Several ATR 42 icing incidents with ice aft of the boots (Air Mauritius, Ryan Air, and Continental Express at Burlington) occurred in layer clouds, which supports the conclusion that icing encounters in high altitude layer clouds can exceed the capabilities of aircraft certified to the Appendix C envelope. Thus, because the Appendix C envelope is limited and does not include larger water drop conditions, such as freezing drizzle or freezing rain (conditions that can be routinely encountered in winter operations throughout much of the northern United States, and were most likely encountered by flight 4184), the Safety Board concludes that the current process by which aircraft are certified using the Appendix C icing envelope is inadequate and does not require manufacturers to sufficiently demonstrate the airplane's capabilities under a sufficiently realistic range of icing conditions. In addition, the lack of standardized methods for processing icing data to determine MVDs raises concern that certification icing tests may be conducted at actual MVDs below the calculated values. For example, during the series of icing tanker tests at Edwards AFB, it was determined that two generally accepted methods of calculating MVD and LWC provided significantly different results. One method was developed by Particle Measuring Systems, the manufacturer of the instruments used to measure the icing conditions, and the other method was developed by NCAR. It was found that when processing any given set of raw icing data collected behind the icing tanker, the two methods provided MVD and LWC results that differed by as much as a factor of 2. These differences are attributed to the different mathematical equations used by the two methods and raise concerns about the accuracy of the results.

[ANALYSIS] Nonetheless, because ATR had indicated that the airplanes in these incidents were inappropriately flown in icing conditions beyond the certification envelope, and that in most cases the pilots had not increased the propeller speed to 86 percent (as required by the aircraft flight manual procedure for flight in icing conditions), it was reasonable for the DGAC to accept ATR's commitment to educate flightcrews with the All Weather Operations brochure as an adequate response at that time. However, the Safety Board concludes that the DGAC did not require ATR to include adequate information about sudden autopilot disconnects, and rapid, uncommanded aileron and control wheel deflections in its All Weather Operation brochure, nor did the DGAC require that ATR flightcrews receive mandatory training on this subject. During its investigation of the 1993 Continental Express incident at Newark, New Jersey, ATR concluded that turbulence was a primary factor in the upset of the airplane. Excessive ice accumulation on the wings was also identified but attributed to freezing rain and the flightcrew's failure to increase the propeller speed from 77 percent to 86 percent as required by the flight manual. However, the DGAC should have known that the amount of turbulence in that incident (+/-0.3 G) was too low a turbulence level to cause aileron hinge moment reversals in a 181 transport category aircraft, and therefore should have recognized that the freezing rain encounter was the reason for the unstable aileron behavior. Further, the 1994 investigation of the Continental Express incident at Burlington, Massachusetts, provided data that led to the conclusion that an ice-induced aileron hinge moment reversal occurred after "severe" ice had caused the airplane to decelerate and pitch up despite proper use of all ice protection procedures. Based on the long history of ATR incidents in icing conditions, especially those that occurred after 1992, the DGAC should have recognized that the vortex generators, the AAS, and the All Weather Operations brochure were not sufficient to correct or prevent the recurrence of the ice-induced aileron hinge moment reversal problem. Further, it should have been clear that the ATR airplanes were still being flown into icing conditions that were beyond the Appendix C envelope or were otherwise conducive to aileron hinge moment reversals. The Safety Board concludes that following the 1994 Burlington incident, the DGAC should have required ATR to take further action to correct the ice-induced aileron instability and to ensure that all operators and regulators were aware of ATR's analyses of the incidents and the characteristics of the phenomenon. Further, the Safety Board concludes that the DGAC's failure to require ATR to take additional corrective actions, such as performing additional icing tests, issuing more specific warnings regarding the aileron hinge moment reversal phenomenon, developing additional airplane modifications, and providing specific guidance on the recovery from a hinge moment reversal, led directly to this accident. 2.6.3 Continuing Airworthiness Oversight by FAA As early as 1981, the Safety Board had recommended that freezing rain be included in the Appendix C envelope because aircraft operate in such conditions.82 In 1983, Dr.

[ANALYSIS] Moreover, in-service ATR incidents and pilot reports have shown that side window icing does not always accompany ice accretions aft of the deice boots, which ATR has stated only occurs in freezing drizzle and/or freezing rain. The Safety Board acknowledges the efforts of atmospheric research in the meteorological community and hopes that its important findings will eventually provide the aviation industry with a better understanding of the freezing drizzle/rain phenomenon. The Safety Board concludes that the continued development of equipment to measure and monitor the atmosphere (i.e., atmospheric profilers, use of the WSR-88D and Terminal Doppler weather radars, multispectral satellite data, aircraft-transmitted atmospheric reports, and sophisticated mesoscale models), and the development of computer algorithms, such as those contained in the FAA's Advanced Weather Products Generator (AWPG) program to provide comprehensive aviation weather warnings, could permit forecasters to refine the data sufficiently to produce more accurate icing forecasts and real-time warnings. Therefore, the Safety Board believes that the FAA should continue to sponsor the development of methods to produce weather forecasts that define very specific locations of potentially hazardous atmospheric icing conditions (including freezing drizzle and freezing rain) and to produce short-range forecasts ("nowcasts") that identify icing conditions for a specific geographic area with a valid time of 2 hours or less. 2.4 ATR Flight Characteristics in Icing Conditions As discussed previously, the evaluation of meteorological data indicates that the range of water droplet sizes at Roselawn probably varied from cloud drops of less than 50 microns to drops as large as 2,000 microns. The 100 to 140 micron MVD drop sizes in the December 1994 ATR icing tanker tests resulted 168 in ice ridges just aft of the active portion of the deice boots and subsequent autopilot and aileron behavior comparable to that noted in the FDR data for the accident. Control wheel force data from the icing tanker tests and the subsequent flight tests with artificial ice shapes indicated that the freezing drizzle ice shapes caused trailing edge flow separation and subsequent aileron hinge moment reversals. Therefore, the Safety Board concludes that the ATR 42 and 72 can experience ice-induced aileron hinge moment reversals, autopilot disconnects, and rapid, uncommanded rolls if they are operated in near-freezing temperatures and water droplet MVDs typical of freezing drizzle. The freezing drizzle encounters in the December 1994 ATR icing tanker tests resulted in ice ridge accretions aft of the deice boots in both the flaps 0 and flaps 15 configurations. However, the tanker test results showed that at flaps 15, there was no pronounced ice ridge on the lower wing surface as there was when the ice accreted at flaps 0. Further, there was a much smaller drag increase when the ice accreted at flaps 15 than there was when the ice accreted at flaps 0. Based on the small drag increases apparent in the data from flight 4184, it is apparent that the ice ridge that formed during the accident flight developed and grew primarily after the flaps were extended to 15 degrees.

[ANALYSIS] However, FSAT 95-29 does not specifically state that continued flight in such conditions is prohibited. The Safety Board is concerned that in some situations it may be necessary to operate in such conditions for an extended period of time. One such situation is the failure of an engine shortly after takeoff, which could require maneuvering for an indeterminate period of time while returning to the departure airport where freezing drizzle or light freezing rain conditions are known to exist. Further, although it is known by many in the aviation community that flight into freezing drizzle or freezing rain is not safe, the Safety Board is unaware of an explicit provision in the Federal Aviation Regulations that prohibits flight into freezing drizzle and freezing rain. Additionally, as was noted in the Safety Board's 1981 study on aircraft icing, airplanes certificated for flight into known icing are 80The National Center for Atmospheric Research (NCAR) definition for light freezing rain is: "measured intensity up to 0.10 in/hr (2.5 mm or 25 gr/dm2/hr); Maximum 0.01 inch in 6 minutes from scattered drops that, regardless of duration, do not completely wet an exposed surface up to a condition where individual drops are easily seen." 173 authorized to fly into weather conditions that produce "severe" icing under 14 CFR Parts 91, 135 and 121. However, by definition, severe icing conditions result in a rate of ice accumulation that exceeds the capabilities of the airplane deice/anti-icing system or that require immediate diversion from the planned route of flight. The Safety Board is concerned that these unclear and inconsistent messages to pilots about the operation of aircraft that are certified for flight in icing conditions may create the misconception that flight in freezing drizzle and/or freezing rain is acceptable when it is not. Such confusing and apparently contradictory information could have contributed to the belief by Simmons Airlines/AMR Eagle management that it was permissible for ATR 42 and 72 airplanes to be dispatched and flown into conditions of freezing drizzle and light freezing rain when it disseminated a memorandum to its pilots in 1991 setting forth the conditions for such flights. The Safety Board concludes that no airplane should be authorized or certified for flight into icing conditions more severe than those to which the airplane was subjected in certification testing unless the manufacturer can otherwise demonstrate the safety of flight in such conditions. Thus, the Safety Board believes that the FAA should revise its certification regulations to ensure that airplanes are properly tested for all conditions in which they are authorized to operate, or are otherwise shown to be capable of safe flight into such conditions. If safe operation cannot be demonstrated by the manufacturer, operational limitations should be imposed to prohibit flight in such conditions, and flightcrews should be provided with the means to positively determine when they are in icing conditions that exceed the limits for aircraft certification.

[ANALYSIS] In addition, during development of the ATR 72-210 series airplanes, ATR added additional vortex generators to further increase the AOA at which the aileron hinge moment reversals occurred. ATR also specifically designed the stall protection system to activate at AOAs lower than the AOA at which the ailerons would become unstable. The knowledge gained from flight testing about hinge moment reversals, the findings from the previous incidents, and ATR’s active 177 participation in the study of tailplane icing hinge moment reversals, leads the Safety Board to conclude that ATR recognized the reason for the aileron behavior in the previous incidents and determined that ice accumulation behind the deice boots, at an AOA sufficient to cause an airflow separation, would cause the ailerons to become unstable. ATR had sufficient basis to modify the airplane and/or provide airworthiness authorities, operators, and pilots with adequate, detailed information regarding this phenomenon. Following the Mosinee incident, ATR issued an “Operators Information Message” (OIM), added the Anti-Icing Advisory System (AAS) and vortex generators in front of the ailerons to all ATR airplanes, proposed airplane flight manual (AFM) and flightcrew operating manual (FCOM) changes and developed an icing package for the ATR simulators. The 1989 OIM characterized the Mosinee incident in a manner that could have been interpreted by operators and pilots to indicate only that the ailerons became “stiff” or hard to move because of an accretion of ice, and that the autopilot was unable to move the ailerons and correct the increasing roll attitude. The OIM did not indicate that an ice accretion behind the deice boots in front of the ailerons, could cause them to overpower the autopilot, move uncommanded in an abrupt manner to their full travel limits, cause rapid rolls and create unusual lateral control behavior. The 1989 icing simulation package provided to simulator manufacturers and aircraft operators by ATR for use in their ATR 42 training programs did not adequately present the effects of the icing event experienced by the Mosinee flightcrew or the crew of flight 4184. The modification did not model abrupt, full aileron and control wheel deflections with high control wheel forces that would typically be necessary to recover from an aileron hinge moment reversal. Further, AMR Eagle stated that in its simulator program, the pilots were taught to initiate recovery from a stall at the first indication of stick shaker, stall aural warning, airframe buffet or stick pusher, and that any training in an incorrect configuration, such as further increasing the AOA beyond stall warning to cause the airplane to roll off, would be classified as "negative" training. In 1996, after the Roselawn accident, ATR provided simulator operators with another, more accurate icing simulation package.

[ANALYSIS] However, the height of the shapes was 1/4 inch, which ATR indicates was not sufficient to initiate an aileron hinge moment reversal prior to SPS activation. Following the two ATR 42 incidents in 1991 (involving Ryan Air and Air Mauritius), which ATR also attributed to operation in icing conditions outside of the certification envelope, ATR published its 1992 All Weather Operations brochure.81 The brochure, which was sent to all ATR operators, provided 81The All Weather Operations brochure, in which ATR consolidated general aircraft operating information for flight in all types of weather conditions (including freezing drizzle and freezing rain), was not provided by Simmons Airlines/AMR Eagle to its pilots. Simmons Airlines/AMR Eagle management stated that the brochure was not disseminated because some information was contrary to Federal regulations and because most of the information already existed in the various approved flight and operating manuals. 179 information about freezing rain, including temperature ranges that could produce such conditions. It stated “Aileron forces are somewhat increased when ice accretion develops, but remain otherwise in the conventional sense,” which is inconsistent with the actual rapid and uncommanded aileron and control wheel deflections to near their full travel limits with unusually high, unstable control wheel forces. The brochure also stated “Freezing rain is capable of rapidly covering an aircraft with a sizable layer of clear ice, well beyond the usual accretion areas around the stagnation point.” However, the statement does not specifically indicate that ice may accumulate a significant distance beyond the deice boots, although the wing leading edges and windscreen may be free of ice. Finally, the brochure stated that “Should the aircraft enter a freezing rain zone, the following procedures should be applied: Autopilot engaged, monitor retrim roll left/right wing down messages. In case of roll axis anomaly, disconnect autopilot holding the control stick firmly.” However, this does not indicate that a roll trim message may not occur, or could occur coincident with the autopilot disconnecting (as it did with flight 4184), thus precluding sufficient time for the flightcrew to perform the recommended procedures, nor does it advise flightcrews to expect sudden autopilot disconnects, rapid and uncommanded aileron and control wheel deflections to near their full travel limits with unusually high, unstable control wheel forces. Therefore, the Safety Board concludes that the ATR All Weather Operations brochure was misleading and minimized the known catastrophic potential of ATR operations in freezing rain.

[ANALYSIS] Therefore, the Safety Board concludes that the ATR All Weather Operations brochure was misleading and minimized the known catastrophic potential of ATR operations in freezing rain. Following the 1993 incident at Newark, New Jersey, which ATR attributed to turbulence and freezing rain, and the incident about 1 year later at Burlington, Massachusetts (both of which involved aileron-hinge moment reversals in icing conditions), ATR had sufficient knowledge to conclude that the ATR 42 had a significant, recurring airworthiness problem in icing conditions outside the Appendix C icing certification envelope. Although ATR knew that the icing conditions encountered in these incidents were outside the icing certification envelope, ATR also knew that the airplanes were being flown more than occasionally into such conditions and that neither the vortex generators nor the operational information it had disseminated had corrected the problem or prevented recurrence. Therefore, the Safety Board concludes that ATR's failure to disseminate adequate warnings and guidance to operators about the adverse characteristics of, 180 and techniques to recover from, ice-induced aileron hinge moment reversal events, and ATR's failure to develop additional airplane modifications, led directly to this accident. 2.6.2 Continuing Airworthiness Oversight by DGAC ATR provided the DGAC (but not the FAA) with copies of all its incident analyses, including the incident at Mosinee, Wisconsin. Thus, the DGAC should have been fully aware that ATR had concluded that the Mosinee and other incident flightcrews had flown their airplanes into icing conditions that were beyond the Appendix C icing certification envelope. The DGAC should have recognized from the ATR analyses that such incidents resulted in unexpected autopilot disconnects, and rapid, uncommanded aileron and control wheel deflections. As discussed in Section 2.6.1, following the 1991 Ryan Air and Air Mauritius incidents, ATR developed its 1992 All Weather Operations brochure in which the aileron behavior was vaguely discussed without directly alerting operators or pilots to the specifics of the prior incidents or providing explicit guidance on how to cope with aileron hinge moment reversals. The DGAC did not require ATR to provide more specific information to operators and pilots, nor did it require ATR to do further research and testing in icing conditions. Nonetheless, because ATR had indicated that the airplanes in these incidents were inappropriately flown in icing conditions beyond the certification envelope, and that in most cases the pilots had not increased the propeller speed to 86 percent (as required by the aircraft flight manual procedure for flight in icing conditions), it was reasonable for the DGAC to accept ATR's commitment to educate flightcrews with the All Weather Operations brochure as an adequate response at that time.

[ANALYSIS] Also, the ridge of ice that formed in the tanker tests and in the NASA-Lewis icing tunnel tests tended to shed pieces randomly along the span of the wing, resulting in broken, jagged ridges. Although these tests only involved exposing a small portion of the outboard section (including the aileron) of one wing to freezing drizzle, it is likely that the random nature of the partial ice shedding would result in airflow asymmetry over the left and right ailerons in a natural encounter of the airplane with freezing drizzle. Such asymmetry could cause an aileron hinge moment reversal. Because the ailerons on the ATR 72 are not hydraulically actuated, a pilot would have to overcome manually the rapid increase in force produced by a hinge moment reversal. For the accident conditions at the initiation of the aileron hinge moment reversal (185 KIAS, 1.0 G), ATR indicates that approximately 60 pounds of force on the control wheel would have been required to maintain a wingslevel-attitude. This amount of force is in compliance with Federal Aviation Regulations for temporary control wheel forces. However, the Safety Board concludes that rapid, uncommanded rolls and sudden multiple onsets of even 60 pounds of control wheel force without any form of warning or pilot training for 169 such unusual events would, and most likely did in this case, preclude the flightcrew from effecting a timely recovery. The Safety Board recognizes that the risk of another ATR 42 or 72 accident resulting from an uncommanded aileron excursion in freezing drizzle/freezing rain has been reduced by the addition of extended deice boots, improved operational procedures, extensive crew training, and heightened awareness by pilots. Because wind tunnel and in-flight tanker tests have been performed for only a limited range of icing and flight conditions, the Safety Board remains concerned whether, even with the improvements, the airplane can be controlled under all naturally occurring combinations of conditions of liquid drop size and content, temperature, airplane configuration, load factors, speeds, and time of exposure. Moreover, the Safety Board found that ATR’s post-Roselawn brochure entitled, “ATR Icing Conditions Procedures,” still does not adequately address or clearly represent the exact nature of the ATR ice-induced aileron hinge moment reversal. Additionally, as part of the investigation, the Safety Board reviewed historical accident and incident data of other similar turbopropeller aircraft. The data did not show other airplane models to have a similar incident/accident history involving uncommanded aileron excursions in the presence of freezing drizzle/freezing rain.

[ANALYSIS] At 1525, as the airplane was entering the holding pattern, the propeller speed was reduced to 77 percent. According to AMR Eagle procedures, this action is consistent with the reduction of anti-icing/deicing systems to Level I, which is appropriate only for flight outside of actual or potential icing conditions. The FDR indicated that at 1540, the Level III ice protection system was activated and the propeller speed was increased to 86 percent. However, FDR data also revealed that subsequently on two occasions during the holding pattern preceding the initial upset, there was evidence of small drag increases that were probably the result of ice accretions on the airplane. The first drag increase occurred at approximately 1533 (about 24 minutes before the upset78) just before the flaps were extended to 15 degrees. The second increase was evident at about 1551 (6 minutes before the upset). It is likely that the airplane intermittently encountered areas of large supercooled drizzle/rain drops while it was holding which contributed to the formation of a ridge of ice on the upper surface of the wing, aft of the wing deice boots, in front of the ailerons. The crew received a clearance to descend to 8,000 feet. At 1557:23, as they were descending at 185 KIAS, the CVR recorded the activation of the aural flap overspeed warning. The flightcrew retracted the flaps in response to the 78The total time from the start of the hold to the upset was about 39 minutes. 160 warning. The FDR data indicate that as the flaps retracted, the autopilot increased the pitch attitude to maintain a preset vertical speed for the descent. As the airplane pitched nose up and the AOA increased through 5 degrees, the airflow in the area of the right aileron began to separate from the wing upper surface because of the ice ridge. As the AOA continued to increase, the airflow separation in the area of the right aileron also increased, causing a reversal of the right aileron hinge moment characteristics. Although the right aileron hinge moment reversal caused the ailerons to deflect rapidly to a right-wing-down (RWD) position, the AOA was not sufficient to activate the stall warning system prior to the aileron deflection. The autopilot could not control the aileron deflection rate, which exceeded that allowed by the autopilot so the autopilot disconnected. Within 0.25 seconds of the autopilot disconnection, the ailerons fully deflected to the RWD position79 and the airplane rolled rapidly to the right until reaching 77 degrees RWD. An immediate nose-down elevator deflection reduced the AOA; and the ailerons were deflected LWD by the flightcrew to counter the right roll.

[CONCLUSIONS > FINDINGS] Prior to the Roselawn accident, ATR recognized the reason for the aileron behavior in the previous incidents and determined that ice accumulation behind the deice boots, at an AOA sufficient to cause an airflow separation, would cause the ailerons to become unstable. Therefore, ATR had sufficient basis to modify the airplane and/or 206 provide operators and pilots with adequate, detailed information regarding this phenomenon. 22. The 1989 icing simulation package developed by ATR for the training simulators did not provide training for pilots to recognize the onset of an aileron hinge moment reversal or to execute the appropriate recovery techniques. 23. ATR’s proposed post-Mosinee AFM/FCOM changes, even if adopted by the DGAC and the FAA, would not have provided flightcrews with sufficient information to identify or recover from the type of event that occurred at Roselawn, and the actions taken by ATR following the Mosinee incident were insufficient. 24. The 1992 ATR All Weather Operations brochure was misleading and minimized the known catastrophic potential of ATR operations in freezing rain. 25. ATR failed to disseminate adequate warnings and guidance to operators about the adverse characteristics of, and techniques to recover from, ice-induced aileron hinge moment reversal events; and ATR failed to develop additional airplane modifications, which led directly to this accident. 26. The DGAC failed to require ATR to take additional corrective actions, such as performing additional icing tests, issuing more specific warnings regarding the aileron hinge moment reversal phenomenon, developing additional airplane modifications, and providing specific guidance on the recovery from a hinge moment reversal, which led directly to this accident. 27. The FAA's failure, following the 1994 Continental Express incident at Burlington, Massachusetts, to require that additional actions be taken to alert operators and pilots to the specific icing-related problems affecting the ATRs, and to require action by the manufacturer to remedy the airplane's propensity for aileron hinge moment reversals in certain icing conditions, contributed to this accident. 207 28. The FAA Aircraft Evaluation Group (AEG) did not receive in a timely manner, from all sources, pertinent documentation (such as the ATR analyses) regarding the previous ATR icing incidents/accidents that could have been used to monitor the continued airworthiness of the airplane. 29. The ability of the FAA's AEG to monitor, on a real-time basis, the continued airworthiness of the ATR airplanes was hampered by the inadequately defined lines of communication, the inadequate means for the AEG to retrieve pertinent airworthiness information, and the DGAC's failure to provide the FAA with critical airworthiness information, because of the DGAC's apparent belief that the information was not required to be provided under the terms of the Bilateral Airworthiness Agreement (BAA). These deficiencies also raise concerns about the scope and effectiveness of the BAA. 30.

[ANALYSIS] Approximately 1.7 seconds later, as the altitude decreased through 1,700 feet, the first officer made an expletive comment, the elevator position and vertical acceleration began to increase rapidly (to more than 3.7 G), and the CVR recorded a loud crunching sound. The CVR and FDR data end 0.5 second later. Analysis of the airplane wreckage indicates that the outboard 10 feet of the left and right wings, as well as the horizontal stabilizer, separated from the airframe at a very low altitude. The first officer's expletive comment occurred when the airplane was descending through 1,700 feet, which was most likely just after the airplane descended through the base of the clouds (the clouds were broken at about 2,100 feet). The Safety Board concludes that both pilots saw the ground, realized their close proximity, nose-down attitude, and high descent rate, and made an additional nose-up elevator input. This elevator input combined with the high airspeed (about 115 KIAS over the certified maximum operating airspeed) resulted in excessive wing loading and the structural failure of the outboard sections of the wings. 2.3 Meteorological Factors 2.3.1 General Based on the analysis of all available data, reports from pilots and evaluations by several atmospheric scientists and researchers, the Safety Board concludes that flight 4184 encountered a mixture of rime and clear airframe icing in 162 supercooled cloud and drizzle/rain drops, while in the holding pattern at the LUCIT intersection. The supercooled drops in the area were estimated to be greater than 100 microns in diameter, with some as large as 2,000 microns. The liquid water content (LWC) was estimated to have varied from less than 0.1 to nearly 1.0 gram per cubic meter. The ambient air temperature in the area of the holding pattern (10,000 feet) was about minus 3 degrees C, with the freezing level between 7,000 and 8,000 feet, and the cloud tops between 19,000 and 30,000 feet. In addition, there were ice crystals present in the atmosphere along the flightpath traversed by flight 4184. The LWC estimates were predicated on the maximum and minimum reflectivity values from the WSR-88D Doppler weather radar located at Romeoville, Illinois (KLOT), the upper air data for the area of the LUCIT intersection, and calculations performed using the Safety Board's ICE4A computer program and a mathematical equation developed by the U.S. Air Force. The drop sizes in the area of the accident were estimated using the WSR-88D radar data.

[ANALYSIS] At 0800, on the day of the accident, the Chicago Air Route Traffic Control Center (ARTCC) Traffic Management Coordinator (TMC) requested that a ground delay program be implemented for aircraft scheduled to land at O'Hare International Airport between the hours of 1200 and 1800 because of the forecast of unfavorable weather conditions. Flight 4184 was released from IND into an area of forecast icing conditions after a 42-minute ground hold with the anticipation that the flight would probably hold en route. The area supervisor at the Chicago ARTCC testified that it is considered an acceptable practice to issue a holding clearance for turbopropeller aircraft operating in "light" or "moderate" icing conditions. Icing conditions often do not exist even though such conditions are forecast. The supervisor stated that the controllers would be "very responsive" if a pilot indicated that they were holding in icing conditions and "wanted to get out," or rejected a holding pattern because of icing conditions. The supervisor also stated that on the day of the accident, he was not aware of any flightcrews rejecting holding instructions because of icing conditions. Because forecasts of hazardous weather may not be precise, and because airplanes can encounter a variety of icing conditions including those considered to be "severe," and exit the conditions safely, efficient use of airports is typically achieved by dispatching aircraft at rates that may require holding if the weather deteriorates. Therefore, the Safety Board concludes that although the controlling facilities were aware that light icing conditions were forecast for the area of the LUCIT intersection, flight 4184 was properly released from Indianapolis because there were viable options for pilots who chose to avoid holding in icing conditions. Periodically, throughout the day of the accident, all sector controllers responsible for aircraft inbound to ORD were advised that the holding of aircraft was possible, and that if the majority of traffic was inbound from the east, the west sectors would hold aircraft, and vice versa. At 1452, when the Chicago Center Flow Control (CFC) advised the Indianapolis Clearance Delivery (CD) that flight 4184 was released, the CD was also told "...that fix is in the hold so he might do some holding when he gets up here...." At 1517, approximately 22 minutes after flight 4184 departed, the sectors responsible for inbound flights from the east [which included flight 4184] were instructed to implement holding because a "rush" of inbound aircraft were arriving from the west sector. FAA Order 7110.65 states that air traffic services will be provided to aircraft on a "...first come first served basis as circumstances permit." However, this is not always feasible during periods of high 189 traffic volume or adverse weather conditions. During these periods, the primary responsibility of traffic management is to ensure the safe and orderly flow of air traffic, which may require the holding of aircraft in some sectors while allowing other aircraft to continue inbound to their destination.

[ANALYSIS] When the BOONE sector controller assumed control of the position and received a briefing by the departing controller, he was told, "...no one was complaining about the weather." This included flight 4184 which had been on the radio frequency for approximately 3 minutes when the BOONE controller assumed control. Because there were no PIREPs provided to 190 the previous controller and the crew of flight 4184 did not provide a PIREP of icing conditions at the LUCIT intersection, it was reasonable for the controller to assume that there were no significant weather events in that area, and that the crew of flight 4184 was not experiencing any problems that would have required the controller to take alternative actions. Nonetheless, the Safety Board believes the FAA should revise FAA Order 7110.65, “Air Traffic Control,” Chapter 2, “General Control,” Section 6, “Weather Information,” paragraph 2-6-3, “PIREP” Information, to include freezing drizzle and freezing rain. These conditions should also be clearly defined in the Pilot/Controller Glossary. 2.9 Flightcrew Actions As noted in section 2.2, flight 4184 entered the holding pattern at 77 percent propeller RPM, which is consistent with the use of Level I ice protection and would have been appropriate only if the flight were operating outside of all clouds and precipitation. However, the drag increase noted at approximately 1533 was evidence of flight in at least intermittent icing conditions in clouds or precipitation. Further, on two occasions while flight 4184 was holding, the CVR recorded a single tone chime identified as a caution alert. The caution alert can be activated by one of several different aircraft systems, including the ice detection system. The flightcrew did not increase the propeller RPM to 86 percent and activate the ice protection system when the first caution alert chime sounded at 1533:56, but following the second caution alert at 1541:07, the FDR indicated that the flightcrew did activate the Level III ice protection system and increased the propeller RPM to 86 percent. Because there was no discussion between the crewmembers regarding the 1533:56 caution alert, it is possible that it was activated by one of the other aircraft systems.

[ANALYSIS] Richard Jeck, now one of the FAA's experts in aircraft icing, raised similar concerns within the FAA. Following the 1988 Mosinee incident, the FAA became aware that the ATR 42 was susceptible to aileron hinge moment reversals in freezing drizzle/light freezing rain conditions. In a 1989 letter to the FAA, the Air Line Pilots Association (ALPA) stated that the AAS and vortex generator modifications to the ATR airplanes were a positive step forward in taking corrective action. However, ALPA questioned whether they were adequate to solve 82See discussion in section 1.18.3. 182 the problem, and also stated its concern that pilots still had no definitive way of identifying when they encounter icing conditions that are outside the certification envelope. The FAA, which had indicated that the vortex generators would correct the aileron anomaly, did not respond to ALPA's concerns except to state that freezing rain is a "...rare, low altitude phenomena, that is generally easy to forecast and therefore avoid[able]." The Safety Board is concerned that the FAA apparently misunderstood the function of the vortex generators when it approved their installation on the ATR 42 and rescinded the previously imposed flight restrictions. The language used in the airworthiness directive (AD) indicated that the FAA believed that the installation of vortex generators would eliminate the aileron hinge moment reversal problem, rather than only delaying the onset of the hinge moment reversal to a greater AOA. The Board is also concerned that apparently neither ATR nor DGAC corrected the FAA's misunderstanding. In 1989, the manager of the FAA's airworthiness evaluation group (AEG) stated in a briefing paper that there had been 10 ATR icing incidents that warranted further study and that "...in the context of problem solving we would like to see flight tests on the ATR series aircraft with irregular shapes emulating "runback" [ice]...." Despite the Safety Board’s request for information, the FAA was unable to provide the Safety Board with a copy of any FAA response to the concerns raised in this paper. Further, in 1991, the FAA led an industry/government team in developing a more detailed understanding of the icing accidents attributed to tailplane icing. Freezing drizzle and freezing rain were a primary topic of discussion. Following the 1991 Ryan Air and Air Mauritius incidents, neither ATR nor DGAC provided the FAA with copies of ATR's analyses of these incidents. Although some FAA staff may have been aware of these incidents and the 1992 ATR All Weather Operations brochure, the FAA may still not have had sufficient information to recognize that the ATR 42's susceptibility to aileron hinge moment reversal required further action by ATR.

[ANALYSIS] It was found that when processing any given set of raw icing data collected behind the icing tanker, the two methods provided MVD and LWC results that differed by as much as a factor of 2. These differences are attributed to the different mathematical equations used by the two methods and raise concerns about the accuracy of the results. Therefore, it is possible that airplanes certificated in accordance with Appendix C criteria may not actually have been tested in the icing conditions described in the certification documentation. Thus, the Safety Board believes that the FAA should revise the icing certification requirements and advisory material to specify the numerical methods to be used in determining median volume diameter (MVD) and liquid water content (LWC) during certification tests. Further, although no aircraft are certified for flight into freezing drizzle or freezing rain, the ATR 72 flight manual did not specify the operational limits and capabilities of the airplane in conditions such as freezing drizzle and freezing rain. Although the "Normal Procedures/Flight Conditions" section of the FAA-approved 172 ATR 42 flight manual (AFM), section 3-02, page 1, dated March 1992, contained the statement, "Operation in freezing rain must be avoided," the "Normal Procedures/Flight Conditions" section of the ATR 72 AFM did not contain the same statement, or any other limitation or prohibition of operation on the ATR 72 in such conditions. At the Safety Board's public hearing, the ATR Vice President, Flight Operations for North America, testified that the omission of this information from the ATR 72 manuals was "not intentional." Currently, FAA ground deice and anti-ice programs permit operators to dispatch aircraft into freezing drizzle and light freezing rain80 provided they use Type II anti-ice fluid and respect the specified holdover timetables. Specifically, Flight Standards Information Bulletin (FSIB) for Air Transport (FSAT), 95-29, dated October 25, 1995, states that Type II deicing fluid will be used when "operating during light freezing rain and freezing drizzle weather conditions" and that the "use of special procedures (i.e. visual inspections, remote deice capability) is required." The Safety Board recognizes that the FAA's intent of this FSAT is to provide operators with the means to dispatch airplanes that will quickly depart and climb through the freezing drizzle or light freezing rain conditions and that the FAA's permission of limited operations in freezing drizzle and light freezing rain is apparently based on the assumption that the airplane will depart within the prescribed "holdover" time of the anti-ice fluid, and transit through the freezing drizzle/light freezing rain conditions with minimal exposure. However, FSAT 95-29 does not specifically state that continued flight in such conditions is prohibited. The Safety Board is concerned that in some situations it may be necessary to operate in such conditions for an extended period of time.

[ANALYSIS] Goodrich impingement study and NASA’s LEWICE program (see section 1.16.2) also predicted a sparse, rough ice accretion aft of the deicing boot on the lower wing surface for some of the tested conditions. However, no ice accretion aft of the deicing boot was noticed during the natural icing certification tests. (See section 2.5.1.) Although it is possible that some of the drag observed in the accident airplane’s performance was the result of a sparse, rough ice accumulation aft of the deicing boot on the lower wing surface, it was not possible to positively determine whether the accident airplane’s ice accretion extended beyond the deicing boot coverage. 145 stitchlines. During tests conducted at a TAT of 26o F, a small, but prominent (½ inch) ridge of ice frequently appeared on the forward portion (0.5 to 1 percent MAC) of the leading edge deicing boot’s upper surface. The IRT test results were used in NASA’s computational studies, which indicated that these pronounced ice ridges tended to act as stall strips, creating more disrupted airflow over the airfoil’s upper surface, further decreasing the lift produced by the airfoil, and resulting in a lower stall AOA than the rough ice accretions alone. NASA’s computational study data indicated that a thin, rough ice accretion with a small, prominent ice ridge can result in a lower stall AOA and a more dramatic drop off/break than the 3-inch ram’s horn ice shape commonly used during initial icing certification testing. The implications of this finding for FAA icing certification criteria are discussed in section 2.6.2. The accident airplane’s performance displayed evidence of adverse effects on both lift and drag during the airplane’s descent to 4,000 feet msl. The degradation exhibited by the accident airplane was consistent with a combination of thin, rough ice accumulation on the impingement area (including both upper and lower wing leading edge surfaces), with possible ice ridge accumulation. Thus, based on its evaluation of the weather, radar, drag information, CVR, existing icing research data, and postaccident icing and wind tunnel test information, the Safety Board concludes that it is likely that Comair flight 3272 gradually accumulated a thin, rough glaze/mixed ice coverage on the leading edge deicing boot surfaces, possibly with ice ridge formation on the leading edge upper surface, as the airplane descended from 7,000 feet msl to 4,000 feet msl in icing conditions; further, this type of ice accretion might have been imperceptible to the pilots.

[ANALYSIS] The IRT observers further noted that IRT lighting conditions and cloud (spray) type greatly affected the conspicuity of the ice accumulation, making it difficult to perceive the ice accumulation during the icing exposure periods. NASA-Lewis’s scientists described the IRT ice accretions as mostly “glaze” ice, like mixed or clear ice in nature, although it looked slightly like rime ice when the IRT was brightly lighted for photographic documentation of the ice accretions because of its roughness. The Safety Board notes that it is possible that such an accumulation would be difficult for pilots to perceive visually during flight, particularly in low light conditions. This type of accumulation would be consistent with the accident airplane’s CVR, which did not 173 Although the Safety Board considered other possible sources for the aerodynamic degradation (such as a mechanical malfunction), the physical evidence did not support a system or structural failure, and the FDR data indicated a gradual, steadily increasing performance degradation that was consistent with degradation observed by the Safety Board in data from events in which icing was a known factor. 174 All pilot reports indicated moderate or less ice accretions, except the pilots of NW flight 272, who reported that they encountered a trace of rime ice during the descent, then encountered moderate-to-severe icing at 4,000 feet msl about 2 minutes after the accident. 175 These TATs are equivalent to SATs of 21o F (-6o C) to 25.5o F (-3o C). 144 record any crew discussion of perceived ice accumulation and/or the need to activate deicing boots during the last 5 minutes of the accident flight. The location of rough ice coverage observed during the icing tunnel tests varied, depending on AOA; at lower AOAs, the ice accretions extended farther aft on the upper wing surface (to the aft edge of the deicing boot on the upper wing surface, about 7 percent of the wing chord at the aileron midspan), whereas at higher AOAs, the ice accretions extended farther aft on the lower wing surface. In some IRT test conditions, sparse feather-type ice accretion extended aft of the deicing boot coverage on the lower wing surface (which extends to about 10½ percent of the airfoil chord at the aileron midspan) as far as 30 to 35 percent of the airfoil’s chord.176 The density of the rough ice coverage also varied, depending on the exposure time; a sparse layer of rough ice usually accreted on the entire impingement area during the first 30 seconds to 1 minute of exposure, and the layer became thicker and more dense as exposure time increased.

[ANALYSIS] In addition, a hazardous situation may develop even if deicing boots are operated throughout an icing encounter as a result of ice accretions on an airplane’s unprotected surfaces, such as aft of the deicing boots. As previously noted, the B.F. Goodrich impingement study, NASA’s LEWICE calculations, and NASA IRT tests indicated that a light accretion may occur on the unprotected lower wing surfaces aft of the deicing boot on the EMB-120. However, Embraer representatives stated that such an ice accretion would result in only a trace of ice accumulating aft of the deicing boots and would have a minimal aerodynamic penalty in drag only. Although there was no evidence of ice accretion aft of the deicing boot during the EMB120 certification natural icing tests and it was not possible to determine whether the accident airplane’s ice accretion extended aft of the deicing boot coverage, it is possible that ice accretion on the unprotected surface aft of the deicing boot could exacerbate a potentially hazardous icing situation. Based on icing and wind tunnel research and information from the Westair incident, the Safety Board concludes that it is possible that ice accretion on unprotected surfaces and intercycle ice accretions on protected surfaces can significantly and adversely affect the aerodynamic performance of an airplane even when leading edge deicing boots are activated and operating normally. Thus, pilots can minimize (but not always prevent) the adverse effects of ice accumulation on the airplane’s leading edges by activating the leading edge deicing boots at the first sign of ice accretion. It is not clear what effect residual ice/ice accretions on unprotected nonleading edge airframe surfaces have on flight handling characteristics. Because not enough is known or understood about icing in general, and especially about the effects of intercycle and residual ice, the Safety Board believes that the FAA should (with NASA and other interested aviation organizations) conduct additional research to identify realistic ice accumulations, to include intercycle and residual ice accumulations and ice accumulations on unprotected surfaces aft of the deicing boots, and to determine the effects and criticality of such ice accumulations; 182 According to the pilots of Westair flight 7233, they were aware that they were operating in “icing conditions;” they stated that they observed ice accumulating on the airplane and had activated the leading edge deicing boots when the airplane entered the clouds during their departure. 153 further, the information developed through such research should be incorporated into aircraft certification requirements and pilot training programs at all levels. The Safety Board considers it likely that future ice detection/protection systems will decrease the hazards associated with icing by incorporating ice detection and protection (automatic activation of deicing boots or anti-icing systems) for individual surfaces, including the horizontal stabilizers, of all airplanes certificated for flight in icing conditions.

[ANALYSIS] Further, as has been recognized for 50 years or more, and demonstrated in accidents in the 1970s, 1980s, and early 1990s, and then again in the Comair flight 3272 accident, surface roughness/ice accretions that may be imperceptible or appear insignificant to 191 The FAA has since required manufacturers of turbopropeller-driven airplanes to develop visual cues for SLD icing; however, the cues were based on very limited testing. Thus, the Safety Board is not convinced that such cues will exist for all icing conditions outside the appendix C icing envelope. 168 pilots can adversely affect the operation of the airplane. However, because of the imperceptible or seemingly insignificant nature of those accretions, pilots who operate the airplane’s deicing boots in accordance with manufacturer’s guidance (that advises them to wait until a recommended thickness of ice accretes) may not activate the deicing boots under these circumstances. An article written by a Douglas Aircraft Company design engineer (published in January 1979) indicated that although most pilots are aware of the adverse aerodynamic effects of large amounts of ice, pilots appear less aware that seemingly insignificant amounts of thin, rough ice on an airfoil’s leading edge can significantly degrade the airplane’s flight characteristics. The deicing boot operating procedures now contained in most airplane manuals contribute to this lack of awareness by advising pilots to wait until a recommended thickness of ice accretes. During the investigation of this accident, arguments were made that the pilots caused the accident because they accepted an airspeed 10 knots slower than Comair’s FSM recommended for holding in icing conditions. However, the Safety Board notes that an EMB120 loaded and configured similar to Comair flight 3272, and operated at 150 knots without any ice accretions, would have a 36-knot margin between its operating airspeed and the stall speed. This margin would likely appear to be an adequate safety margin to a pilot who did not recognize that the airplane was accumulating ice or did not believe that enough ice had accumulated to warrant activation of the deicing boots. The flight handling testing that occurred during the icing certification process did not identify that control problems that were observed in the accident airplane’s performance at an airspeed of about 156 knots (only 4 knots below the 160-knot minimum speed for flight in icing conditions set by the FAA following the Comair accident) with only a small amount of ice accreted on the deicing boots.

[ANALYSIS] According to the NCAR report, the area of reduced reflectivity indicated that “the snow-making process was less efficient there, thus allowing a greater opportunity for liquid cloud to exist.” Postaccident statements obtained from the other pilots who were operating along the accident airplane’s flightpath (and passed through the area of low reflectivity) near the time of the accident indicated that they encountered widely variable conditions. For example, the pilots of Cactus 50 reported moderate rime icing with the possibility of freezing drizzle, the pilots of NW flight 272 encountered moderate-to-severe rime icing as soon as they leveled off at 4,000 feet msl, and the pilots of NW flight 483 reported no icing. Comparison of data from the airplanes indicates that the differences in airframe ice accretion reported by the pilots can be attributed to slight differences in timing, altitude, location (ground track), airspeed, and icing exposure time (and time within the area of reduced reflectivity) of the airplanes. Based on weather radar information and pilot statements, the Safety Board concludes that the weather conditions near the accident site were highly variable and were conducive to the formation of rime or mixed ice at various altitudes and in various amounts, rates, and types of accumulation; if SLD icing conditions were present, the droplet sizes probably did not exceed 400 microns and most likely existed near 4,000 feet msl. 2.4 Aerodynamic Effect of the Ice Accretion on Comair Flight 3272 To help assess the type, amount, and effect of the ice that might have been accumulated by Comair flight 3272 during its descent, the Safety Board reviewed the available icing and wind tunnel research data, conducted additional airplane performance studies/simulations, and requested NASA’s assistance in conducting icing research tunnel (IRT) 143 tests and computational studies.

[CONCLUSIONS > FINDINGS] Because the pilots of Comair flight 3272 were operating the airplane with the autopilot engaged during a series of descents, right and left turns, power adjustments, and airspeed reductions, they might not have perceived the airplane’s gradually deteriorating performance. 11. The accident airplane’s left roll tendency was precipitated by a thin layer of rough ice that accumulated on the leading edge of the wing during the airplane’s cruise descent, and was then affected by some or all of the following factors: the autopilot-commanded left roll, asymmetrical ice selfshedding, aileron deflection effects (localized airflow separations), the effects of engine/propeller thrust, the asymmetrical power application, and the disengagement of the autopilot. It is unlikely that the absence of conductive edge sealer on the left wing leading edge deicing boot segments was a factor in the airplane’s excessive left roll. 12. Consistent with Comair’s procedures regarding ice protection systems, the pilots did not activate the leading edge deicing boots during their descent and approach to the Detroit area, likely because they did not perceive that the airplane was accreting significant (if any) structural ice. 13. Had the pilots of Comair flight 3272 been aware of the specific airspeed, configuration, and icing circumstances of the six previous EMB-120 icingrelated events and of the information contained in operational bulletin 120002/96 and revision 43 to the EMB-120 airplane flight manual, it is possible that they would have operated the airplane more conservatively with regard to airspeed and flap configuration or activated the deicing boots when they knew they were in icing conditions. 14. The current operating procedures recommending that pilots wait until ice accumulates to an observable thickness before activating leading edge deicing boots results in unnecessary exposure to a significant risk for turbopropeller-driven airplane flight operations. Based primarily on concerns about ice bridging, pilots continue to use procedures and practices that increase the likelihood of (potentially hazardous) degraded airplane performance resulting from small amounts of rough ice accumulated on the leading edges. 15. It is possible that ice accretion on unprotected surfaces and intercycle ice accretions on protected surfaces can significantly and adversely affect the 178 aerodynamic performance of an airplane even when leading edge deicing boots are activated and operating normally. 16. Current ice detection/protection requirements and application of technology (particularly deice boots) may not provide adequate protection for a variety of ice accumulation scenarios (tailplane, supercooled large droplets, thin, rough ice accumulations, etc.). 17. The guidance provided by Comair in its memos, bulletins, manuals, and training program did not adequately communicate or emphasize specific minimum airspeeds for operating the EMB-120 in the flaps-up configuration, in or out of icing conditions, and thus contributed to the accident. 18.

[ANALYSIS] Further, icing of the magnitude described by the pilots of NW flight 272 would have produced strong visual cues, and it is likely that the pilots would have commented on such a rapid accumulation, had it occurred. As previously noted, the accident airplane’s CVR did not record any flightcrew comments about ice accumulation or the need to activate the leading edge deicing boots during the last 5 minutes of the accident flight; this is consistent with an ice accumulation that was either not observed by the pilots or that was observed, but considered to be unremarkable. 2.4.1 Possible Factors in Left Roll Tendency Although the accident airplane’s entire airframe was exposed to roughly equivalent icing conditions, making it theoretically possible that ice accumulation would be symmetrical, icing research and wind tunnel tests revealed that ice accumulation (especially ice accumulated at near freezing temperatures, as occurred in this accident) is rarely totally symmetrical, either physically or aerodynamically. FDR information showed that, from the time the airplane leveled at 4,000 feet msl, and then began the left turn, the airplane exhibited an increasing left roll tendency. The Safety Board identified several factors that might have contributed in varying degrees to the left roll, asymmetric conditions, and loss of control observed in this accident. These possible factors included the following: • asymmetrical ice accumulation effects, possibly caused by ice self-shedding. NASA’s IRT tests indicated that the TAT present at the time of the accident was likely to result in asymmetrical ice self-shedding. Increased vertical acceleration values might have exacerbated the ice self-shedding because of wing bending. • the aerodynamic effects of aileron deflections on airfoils with an ice contaminated leading edge (after the left bank was established, the autopilot commanded right-wing-down aileron deflections—left aileron down, right aileron up—to resist the steepening left roll) during the left turn. Computational two-dimensional studies and wind tunnel research indicated that at higher AOAs the downward aileron deflection could initiate a localized flow separation, which resulted in a decrease in lift on the left wing. The 177 Results from the SLD icing tanker tests suggest that the visual cues for SLD ice accumulations (unusually extensive ice accreted on the airframe in areas not normally observed to collect ice, accumulation of ice on the upper surface of the wing aft of the protected area, and on the propeller spinner farther aft than normally observed) would have been very apparent to the pilots and might have resulted in a comment. 147 localized flow separation occurred at a lower AOA on the downward deflecting aileron than the upward deflecting aileron, thereby considerably reducing (or even reversing)178 the rolling moment induced by the wheel inputs.

[ANALYSIS] Many of the concerns raised about icing in this investigation were previously identified by the Safety Board as early as its September 1981 study on icing avoidance and protection. The study raised concerns about the adequacy of the Part 25 appendix C envelope and icing certification and the difficulties in defining and forecasting icing conditions; as a result of the study, the Safety Board recommended, in part, that the FAA evaluate individual aircraft 165 performance in icing conditions and establish operational limits, review icing criteria in Part 25 and expand (adjust) the Part 25 appendix C envelope as necessary, and establish standardized procedures for icing certification. For many years, the FAA did not respond positively to the Safety Board’s recommendations, indicating that icing was not a significant problem for airplanes certificated under Part 25 appendix C. However, subsequent icing-related accidents at Pasco, Washington (in December 1989), and Beckley, West Virginia (in January 1991), revealed that flight control anomalies could result from tailplane icing (see section 1.18.7) and an icingrelated accident at Cleveland, Ohio (in February 1991), revealed that slightly rough ice accumulations on the wing upper surface can result in hazardous flight handling characteristics.188 Further, the October 1994 ATR-72 accident at Roselawn, Indiana, demonstrated that icing outside the Part 25 appendix C envelope could be a significant problem for airplanes certificated to operate in icing conditions. After this series of fatal accidents (all of which involved icing in transport airplanes operated in air carrier service) drew attention to icing-related hazards, the FAA reacted incrementally to tailplane icing, then rough ice accumulations on the upper wing, and then, later, to runback icing (SLD). The Safety Board recognizes that following the Comair flight 3272 accident, the FAA began an important icing-related research program with Embraer and the UIUC. This work has resulted in findings about the effects of thin/rough ice accretions and ice ridges on boots, with other possible factors (such as intercycle icing and residual ice on boots) as yet unknown or unresolved. However, had the FAA adequately responded to the Safety Board’s 1981 icing recommendation, the earlier accidents, or the concerns expressed in its own staff’s draft report on the EMB-120 and conducted a thorough program of icing-related research that defined a course of action to prevent similar incidents by addressing the certification and operational issues (autopilot use in icing conditions, no autopilot bank angle exceedence warning, no stall warning/protection system adjustment for icing conditions, the effects of thin, rough ice and SLD accretions, etc.), this accident would likely have been avoided.

[ANALYSIS] The Safety Board notes that the failure of the FAA to promptly and systematically address these certification and operational issues resulted in the pilots of Comair flight 3272 being in a situation in which they lacked sufficient tools (autopilot bank angle warning, adjusted stall warning/protection system, ice detection system, adequate deice procedures) and information (airspeed guidance, hazards of thin rough ice accretions, and absence of ice bridging) to operate safely. The Safety Board concludes that despite the accumulated lessons of several major accidents and (in the case of the EMB-120) the specific findings of a staff engineer, the FAA failed to adopt a systematic and proactive (rather than incremental and reactive) approach to the certification and operational issues of turbopropeller-driven transport airplane icing, which was causal to this accident. 188 As discussed in section 1.18.1.1, there have been five DC-9 series 10 airplane takeoff accidents attributed to upper wing ice contamination in the United States since 1968. Although these accidents involved turbojet-driven airplanes (not turbopropeller-driven airplanes, like the other icing-related incidents/accidents discussed in this report), the issue of the FAA’s failure to address icing-related operational and certification issues is pertinent to all airplanes certificated for flight in icing conditions. 166 2.6.2 Icing Certification Requirements The Safety Board reviewed EMB-120 test data from the original certification of the airplane for flight in icing conditions (U.S. and Canadian tests) and the subsequent SLD icing certification tests, which were conducted in 1995 as a result of the ATR-72 accident near Roselawn, Indiana. The Safety Board found no evidence that the EMB-120 did not satisfy the tests to which it was subjected; in fact, during these tests, Embraer demonstrated the airplane’s flight handling qualities under conditions that exceeded the boundaries of the Part 25 appendix C envelope in terms of LWC. Despite the apparent fulfillment of all icing certification requirements by the EMB-120, Comair flight 3272 crashed after apparently accreting a thin layer of rough, “sandpaper-type” ice, in icing conditions that likely fell mostly within the boundaries of Part 25 appendix C, although droplets as large as 400 microns might have been present. Consequently, the Safety Board reviewed the adequacy of the current FAA requirements for the certification of airplanes for flight in icing conditions. For an airplane to be certificated for flight in icing conditions, the FAA requires the manufacturer to demonstrate a limited number of test data points within the Part 25 appendix C envelope. The FAA’s icing certification requirements are based on fully functioning and operating anti-icing and deicing systems.

[ANALYSIS] An FAA engineer reviewed these six incidents in a draft report dated January 26, 1996 (see section 1.18.2.3). The Safety Board has been unable to obtain information about the specific disposition of the draft report within the FAA, although the FAA asserted after the accident that this report did not reflect the official views of the FAA. Nevertheless, the Safety Board notes that more than 1 year before the accident, at least some members of the FAA certification staff responsible for handling EMB-120 icing issues were concerned about, and were considering recommendations on, the following issues: 1) the airplane’s roll behavior with ice accretion, 2) high drag from ice accretions that are not considered by the flightcrew to warrant activating the deicing boots, 3) inadequate stall warning in icing conditions, 4) inadequate stall margin with the airspeed established for use in icing conditions, and 5) problems stemming from the use of the autopilot in these conditions. The FAA’s official response to the six preaccident EMB-120 icing-related events, as expressed to the Safety Board by aircraft certification office (ACO) personnel, was that these incidents shared a common factor—flightcrew failure to activate the leading edge deicing boots. The FAA apparently believed that the EMB-120 was safe to operate in icing conditions as long as the boots were operated. Hence, the FAA’s primary action regarding EMB-120 icing before the accident was to approve the Embraer-proposed, CTA-approved revision to the AFM that pilots activate the boots at the first indication of ice accumulation (revision 43). In doing so, the FAA ACO apparently did not accept the draft report’s conclusions, which recognized that pilots would not activate the boots if they did not recognize ice accumulation, that an engaged autopilot masked the tactile cues of icing, and that under these conditions, the flightcrew also could be deprived of an adequate stall warning. The Safety Board notes with disappointment that this was the latest in a series of limited actions taken by the FAA to address the problems of structural icing in transport airplane certification and operation. Basic knowledge about the aerodynamics of icing (including the knowledge regarding the hazards of small amounts of surface roughness/ice) has been well established for the past 50 years (see section 1.18.1), and there is nothing that has been learned in the most recent, postaccident wind tunnel tests and analyses that could not have been learned before this Comair accident. Many of the concerns raised about icing in this investigation were previously identified by the Safety Board as early as its September 1981 study on icing avoidance and protection.

[ANALYSIS] The Safety Board’s review of data from natural icing flight tests revealed that the airplane’s handling characteristics were evaluated with ½-inch accretions on protected surfaces and that the deicing boots’ ability to remove ice accretions of up to ½ inch was assessed. Embraer was not required to demonstrate the EMB-120’s stall characteristics in adverse operational scenarios, including delayed boot activation, intercycle ice accretion, or residual ice on boots. As a result of the existing icing certification procedures, the FAA did not account for a thin ice accumulation (as was identified during this investigation, and which may not be observed or perceived by pilots to be a threat) that could result in a more hazardous situation than the 3-inch ram’s horn shape (which is readily recognizable by pilots as a hazard and would certainly prompt activation of the boots). The Safety Board is concerned that there may be other unaccounted for ice shapes and/or accretion patterns that could result in potentially hazardous performance degradation. The Safety Board is also concerned that the current icing certification process is overly dependent upon pilot performance; the FAA has long based its icing certification policies and practices on the assumption that pilots will perform their duties without error or misperception. FAA icing-related publications indicate that if ice formations other than those considered in the certification process are present, the airplane’s airworthiness may be compromised. After an airplane is certificated by the FAA for flight in appendix C icing conditions, it becomes primarily the pilots’ responsibility to ensure that the airplane is operated in icing conditions for which it was certificated. However, as noted during the investigation of the ATR-72 accident at Roselawn, during normal flight operations, pilots often cannot tell the difference between icing conditions that fall within the appendix C envelope and icing conditions outside the appendix C envelope.191 (For example, a pilot cannot differentiate between 40 micron droplets and 100 micron droplets.) Because pilots often cannot determine whether icing conditions are consistent with “those considered in the certification process” (i.e., limited points within the appendix C certification envelope), or not (i.e., SLD icing conditions, or other potentially hazardous conditions that were not subjected to testing, analysis, or demonstration during icing certification work), it is virtually inevitable that the airplane will unknowingly be operated in icing conditions that fall outside the certification envelope, or in which the airplane had not demonstrated that it could operate safely.

[ANALYSIS] The NASA-Lewis and FAA/UIUC tests indicated that thin, rough ice accretions located on the leading edge and lower surface of the airfoil primarily resulted in increases in drag, whereas thin, rough ice accretions located on the leading edge and upper wing surface had an adverse effect on both lift and drag; this is consistent with information that has been obtained during NACA/NASA icing research conducted since the late 1930s. Data from research conducted in the 1940s and 1950s indicate that an airfoil’s performance can be significantly affected by even a relatively small amount of ice accumulated on the leading edge area, if that accumulation has a rough, sandpaper-type surface. Consistent with these data, NASA’s drag calculations indicated that the thin, rough layer of sandpaper-type ice accumulation resulted in significant drag and lift degradation on the EMB-120 wing section. Further, the thin rough ice accumulation resulted in a decrease in stall AOA similar to that observed in wind tunnel tests with 3-inch ram’s horn ice shapes on protected surfaces and frequently demonstrated a more drastic drop off/break at the stall AOA. FAA/UIUC conducted wind tunnel tests using generic shapes to represent the sandpaper-type roughness with ridges placed on the upper wing surface at 6 percent of the wing chord (farther aft than the ice ridges observed during NASA’s IRT tests); these tests further demonstrated that the ridge type of ice accretion resulted in more adverse aerodynamic effect than the 3-inch ram’s horn ice shapes. As previously noted, NASA’s IRT tests indicated that when an EMB-120 wing is exposed to conditions similar to those encountered by Comair flight 3272 before the accident, the airfoil tended to accrete a small ice ridge (or ridges) along the deicing boot tube segment 176 According to NASA-Lewis scientists, some of the frost accretion observed aft of the deicing boot on the lower wing surface during the icing tunnel tests might have been an artifact of the icing research tunnel (resulting from the higher turbulence, humidity, and heat transfer characteristics of the tunnel); however, the B.F. Goodrich impingement study and NASA’s LEWICE program (see section 1.16.2) also predicted a sparse, rough ice accretion aft of the deicing boot on the lower wing surface for some of the tested conditions.

[ANALYSIS] The flight handling testing that occurred during the icing certification process did not identify that control problems that were observed in the accident airplane’s performance at an airspeed of about 156 knots (only 4 knots below the 160-knot minimum speed for flight in icing conditions set by the FAA following the Comair accident) with only a small amount of ice accreted on the deicing boots. It is possible that if the FAA had required manufacturers to conduct tests with small amounts of rough-textured ice accreted on the protected surfaces (as might occur before boot activation and between boot cycles) during icing certification testing, the absence of an adequate safety margin above the stall speed would have been identified. Further, the FAA could have ensured pilot awareness of icing and adequate stall warning by requiring manufacturers to install ice detectors192 and stall warning systems with reduced AOA thresholds for operations in icing conditions. Based on its concerns that the current icing certification standards did not require testing for all realistic hazardous ice accretion scenarios, in its 1981 icing-related safety study, the Safety Board recommended that the FAA review the adequacy of the 1950s-era Part 25 appendix C icing envelope, update the procedures for aircraft icing certification , and oversee the manufacturers’ evaluations of aircraft performance in various icing conditions. The circumstances of the Comair flight 3272 accident demonstrated again the continuing need for these FAA actions. The Safety Board considers the information that has been available regarding thin, rough ice accretions sufficient to have prompted the FAA to require additional testing within the appendix C envelope to demonstrate the effects of thin, rough ice as part of the icing certification process. Had the FAA required such additional testing, the resultant information regarding the stall margin and operational envelope of the EMB-120 might have been used to define minimum airspeeds for operating the airplane in icing conditions. Therefore, based on its 192 Rosemount ice detectors were first used in military and transport-category airplanes in the early 1970s. 169 review of the history of icing information, the icing-related incident and accident history, the EMB-120 initial icing certification data, the EMB-120 SLD icing controllability test results, and the circumstances of this accident, the Safety Board concludes that the icing certification process has been inadequate because it has not required manufacturers to demonstrate the airplane’s flight handling and stall characteristics under a sufficiently realistic range of adverse ice accretion/flight handling conditions. As a result of its investigation of the 1994 Roselawn accident, the Safety Board issued Safety Recommendations A-96-54 and A-96-56 (currently classified “Open—Acceptable Response”), which, respectively, stated that the FAA should do the following: Revise the icing criteria published in 14 CFR Parts 23 and 25, in light of both recent research into aircraft ice accretion under varying conditions of liquid water content, drop size distribution, and temperature, and recent developments in both the design and use of aircraft. Also, expand the Appendix C icing certification envelope to include freezing drizzle/freezing rain and mixed water/ice crystal conditions as necessary.

[CONCLUSIONS > FINDINGS] The guidance provided by Comair in its memos, bulletins, manuals, and training program did not adequately communicate or emphasize specific minimum airspeeds for operating the EMB-120 in the flaps-up configuration, in or out of icing conditions, and thus contributed to the accident. 18. The pilots likely did not recognize the need to abide by special restrictions on airspeeds that were established for icing conditions because they did not perceive the significance (or presence) of Comair flight 3272’s ice accumulation. 19. Whether the pilots perceived ice accumulating on the airplane or not, they should have recognized that operating in icing conditions at the air traffic control-assigned airspeed of 150 knots with flaps retracted could result in an unsafe flight situation; therefore, their acceptance of the 150-knot airspeed assignment in icing conditions without extending flaps contributed to the accident. 20. Minimum airspeed information for various flap configurations and phases and conditions of flight would be helpful to pilots of all passenger-carrying airplanes. 21. The stall warning system installed in the accident airplane did not provide an adequate warning to the pilots because ice contamination was present on the airplane’s airfoils, and the system was not designed to account for aerodynamic degradation or adjust its warning to compensate for the reduced stall warning margin caused by the ice. 22. The accident airplane’s autopilot was capable of normal operation and appeared to be operating normally during the last minutes of the accident flight, and the autopilot disconnect and warning systems operated in a manner consistent with their design logic. 23. Had the pilots been flying the airplane manually (without the autopilot engaged) they likely would have noted the increased right-wing-down control wheel force needed to maintain the desired left bank, become aware of the airplane’s altered performance characteristics, and increased their 179 airspeed or otherwise altered their flight situation to avoid the loss of control. 24. Disengagement of the autopilot during all operations in icing conditions is necessary to enable pilots to sense the aerodynamic effects of icing and enhance their ability to retain control of the airplane. 25. If the pilots of Comair flight 3272 had received a ground proximity warning system, autopilot, or other system-generated cockpit warning when the airplane first exceeded the autopilot’s maximum bank command limits with the autopilot activated, they might have been able to avoid the unusual attitude condition that resulted from the autopilot’s sudden disengagement. 26. Despite the accumulated lessons of several major accidents and (in the case of the EMB-120) the specific findings of a staff engineer, the Federal Aviation Administration failed to adopt a systematic and proactive (rather than incremental and reactive) approach to the certification and operational issues of turbopropeller-driven transport airplane icing, which was causal to this accident. 27. The icing certification process has been inadequate because it has not required manufacturers to demonstrate the airplane’s flight handling and stall characteristics under a sufficiently realistic range of adverse ice accretion/flight handling conditions. 28.

[ANALYSIS] The left wing leading edge deicing boot segments were most recently replaced in July 1996—it strains credibility to presume that the pilots and mechanics who examined the airplane’s leading edges during preflight and maintenance inspections between July 1996 and the day of the accident did not observe inconsistently applied conductive edge sealer. Further, during postaccident interviews, Comair’s maintenance personnel appeared to be very familiar with the conductive edge sealer application. 148 whether the lack of conductive edge sealer on the upper wing surface at the aft edge of the leading edge deicing boot had potential to act as a preferred ice accumulation location. The NASA-Lewis scientists stated that it was unlikely because there were no perceivable tactile differences (gaps, edges, roughness, etc.) between the leading edge deicing boot and the wing skin to trigger ice accumulation. More importantly, the NASA-Lewis scientists stated that IRT tests had demonstrated that ice accretion that far aft on the upper wing surface would be unlikely to occur in a non-SLD icing environment—it would be more likely to occur in larger SLD droplet sizes, because of the resultant runback and secondary ice accumulation. The Safety Board concludes that the accident airplane’s left roll tendency was precipitated by a thin layer of rough ice that accumulated on the leading edge of the wing during the airplane’s cruise descent and was then affected by some or all of the following factors: the autopilot-commanded left roll, asymmetrical ice self-shedding, aileron deflection effects (localized airflow separations), the effects of engine/propeller thrust, the asymmetrical engine power application, and the disengagement of the autopilot. It is unlikely that the absence of conductive edge sealer on the left wing leading edge deicing boot segments was a factor in the airplane’s excessive left roll. 2.5 Flightcrew Actions 2.5.1 Use of Deice/Anti-ice Equipment The Safety Board attempted to determine whether the airplane’s ice protection systems were operated during the accident airplane’s descent and approach to Detroit Metropolitan/Wayne County Airport (DTW). CVR information showed that when the pilots performed the descent checklist at 1547, they confirmed that the airplane’s “standard seven” anti-ice systems were activated and activated the windshield heat and the propeller deice system.180 This was consistent with guidance contained in Comair’s EMB-120 Flight Standards Manual (FSM), which stated that anti-ice systems should be activated “before flying into known icing conditions” to prevent ice accumulation on the affected surfaces.

[ANALYSIS] In addition, the Safety Board reviewed wind tunnel test data obtained during research conducted by the FAA at the University of Illinois at Urbana/Champaign (UIUC). The Safety Board’s study of the accident airplane’s aerodynamic performance indicated that it began to degrade from ice accumulation173 about 4½ to 5 minutes before the autopilot disengaged, as the airplane descended through 7,000 feet msl; the amount of degradation increased gradually as the airplane descended to 4,000 feet msl. (See section 1.16.1.2.) Based on this gradual performance degradation, weather radar data that showed light precipitation intensities, pilot reports of moderate or less ice accretions,174 and the Safety Board and NCAR weather studies, it appeared likely that Comair flight 3272 encountered icing conditions that fell within the Part 25 appendix C envelope (see section 1.6.1) and/or the lower portion of the SLD icing range during its descent to 4,000 feet msl. Thus, the postaccident icing tunnel tests were performed using LWCs between 0.52 and 0.85 grams per cubic meter and water droplet sizes between 20 microns and 270 microns. Total air temperatures (TAT) used in the icing tunnel tests ranged between 26o F and 31o F (-3o C and -0.5o C),175 consistent with the static air temperature (SAT) values recorded by the FDR during the airplane’s descent from 7,000 to 4,000 feet msl. The exposure time used in the icing tunnel tests was 5 minutes; additional runs were conducted under some test conditions to determine the effect that deicing boot activation had on cleaning the leading edge and on subsequent ice accretions. The icing tunnel tests did not result in thick ice accumulation under any test condition (including SLD droplets); rather, the tests consistently resulted in a thin (0.25 inch accumulation or less), rough “sandpaper-type” ice coverage over a large portion of the airfoil’s leading edge deicing boot surface area (and aft of the deicing boot on the lower wing surface in some test conditions). In addition, in many IRT test conditions, small (½ inch) ice ridges accreted along the leading edge deicing boot seams. (The effects of these ridges will be discussed later in this section.) According to NASA and Safety Board IRT test observers, the thin, rough ice coverages (and ice ridges, where applicable) that accreted on the EMB-120 wing were somewhat translucent and were often difficult to perceive from the observation window.

[ANALYSIS] The Safety Board notes that FAA Order 7110.10L, “Flight Services,” contains a definition of “trace” ice accumulations, that states, in part, “A trace of ice is when ice becomes perceptible….It is not hazardous even though deicing/anti-icing equipment is not utilized unless encountered for an extended period of time [over 1 hour].” Information obtained during this investigation, which echoed the results of research conducted in the 1930s and 1940s, indicated that thin, rough amounts of ice, even in trace amounts can result in hazardous flight conditions. The Safety Board concludes that the suggestion in current FAA publications that “trace” icing is “not hazardous” can mislead pilots and operators about the adverse effects of thin, rough ice accretions. Therefore, the Safety Board believes that the FAA should amend the definition of trace ice contained in FAA Order 7110.10L (and in other FAA documents as applicable) so that it does not indicate that trace icing is not hazardous. The Safety Board notes that in some icing exposure scenarios, pilots could become aware of the performance degradation without observing a significant accumulation of ice on the airplane by observing other cues, such as a decrease in airspeed, excessive pitch trim usage, a higher-than-normal amount of engine power needed to maintain a stabilized condition, and/or anomalous rates of climb or descent. However, the Safety Board concludes that because the pilots of Comair flight 3272 were operating the airplane with the autopilot engaged during a series of descents, right and left turns, power adjustments, and airspeed reductions, they might not have perceived the airplane’s gradually deteriorating performance. 146 Further, although it is possible (based on the icing reported by the pilots of NW flight 272 and the NCAR scientist’s estimation of the likely droplet size distribution in the clouds) that the accident flight encountered SLD icing177 as it reached 4,000 feet msl, the airplane was only at that altitude for about 25 seconds before the upset occurred; during most of that 25 seconds, the FDR data showed that the autopilot was countering the increasing left roll tendency and a sideslip condition was developing. However, even if the accident flight had accumulated ice at the rapid rate reported by the pilots of NW flight 272 (about ½ inch per minute), the accident flight could not have accumulated a large amount of ice during the brief period of time it spent at 4,000 feet before the autopilot disengaged and the loss of control occurred. Further, icing of the magnitude described by the pilots of NW flight 272 would have produced strong visual cues, and it is likely that the pilots would have commented on such a rapid accumulation, had it occurred.

[ANALYSIS] The Safety Board considers it likely that future ice detection/protection systems will decrease the hazards associated with icing by incorporating ice detection and protection (automatic activation of deicing boots or anti-icing systems) for individual surfaces, including the horizontal stabilizers, of all airplanes certificated for flight in icing conditions. However, because ice accretions and their effects are not yet fully understood, the Safety Board concludes that current ice detection/protection requirements and application of technology (particularly deice boots) may not provide adequate protection for a variety of ice accumulation scenarios (tailplane, SLD, thin, rough ice accumulations, etc.). Therefore, the Safety Board believes that the FAA should actively pursue research with airframe manufacturers and other industry personnel to develop effective ice detection/protection systems that will keep critical airplane surfaces free of ice; then require their installation on newly manufactured and in-service airplanes certificated for flight in icing conditions. 2.5.2 Airspeed and Flap Configuration Information Simulator studies conducted during the investigation revealed that the accident airplane’s decreasing airspeed in icing conditions was critical in the development of the accident scenario. According to FDR data, the airplane began to exhibit signs of departure from controlled flight as it decelerated from 155 to 156 knots. Because the pilots accepted an ATC instruction to slow to 150 knots and maintained a flaps-up configuration, the Safety Board evaluated the guidance that Comair provided to its EMB-120 pilots on minimum airspeed in the flaps-up configuration, the Comair flight 3272 flightcrew’s acceptance of this airspeed without adjusting the airplane’s configuration, and the FAA’s requirements for airplane manufacturers with regard to minimum airspeeds. 2.5.2.1 Comair’s Airspeed Guidance During postaccident interviews, some of Comair’s pilot training personnel indicated that the company’s EMB-120 pilot training emphasized the 160-knot minimum airspeed for operating in icing conditions, and Comair’s EMB-120 Program Manager told Safety Board investigators that 170 knots is the only airspeed the company supports for operating with the landing gear and flaps retracted. Although the Safety Board’s review of the airspeed guidance contained in Comair’s EMB-120 FSM revealed that it did not contain specific minimum maneuvering airspeeds for flight in icing conditions and for various airplane configurations, it did contain general airspeed information in descriptions of normal and nonnormal procedures and maneuvers. For example, the technique outlined in Comair’s FSM for an instrument landing system (ILS) approach associated the base leg vector position (which was the accident airplane’s approximate position on the approach before the upset, albeit still about 20 miles from the destination airport) with 170 knots and the flaps 15 configuration.

[CONCLUSIONS > FINDINGS] 3.1 Findings 1. The pilots were properly qualified and certificated to perform the flight during which the accident occurred, and each crewmember had received the training and off-duty time prescribed by the Federal regulations. There was no evidence of any preexisting medical or behavioral conditions that might have adversely affected the flightcrew’s performance. 2. The airplane was certificated, equipped, and dispatched in accordance with Federal regulations and approved Comair procedures. There was no evidence of preexisting mechanical malfunction or other failure of the airplane structure, flight control or other systems, powerplants or propellers that would have contributed to the accident. 3. It is likely that the leading edge deicing system was capable of normal operation during the accident flight. 4. The Detroit terminal radar approach controllers who were involved with flight 3272 were properly qualified and certificated. A review of air traffic control and facility procedures revealed that the controllers followed applicable air traffic and wake turbulence separation rules, and air traffic separation was assured during flight 3272’s approach to the runway. 5. Although the radar ground tracks of Cactus 50 and Comair flight 3272 converged near the accident site, the Safety Board’s review of winds aloft and wake vortex sink rates indicated that Cactus 50’s wake vortices would have been above and northeast of Comair flight 3272’s flightpath near the upset location. Thus, Comair flight 3272 was separated from the vortices vertically and horizontally, and, therefore, wake turbulence was not a factor in the accident. 6. The airplane was aerodynamically clean, with no effective ice accreted, when it began its descent to the Detroit area. 7. The weather conditions near the accident site were highly variable and were conducive to the formation of rime or mixed ice at various altitudes and in various amounts, rates, and types of accumulation; if supercooled large droplet icing conditions were present, the droplet sizes probably did not exceed 400 microns and most likely existed near 4,000 feet mean sea level. 8. It is likely that Comair flight 3272 gradually accumulated a thin, rough glaze/mixed ice coverage on the leading edge deicing boot surfaces, possibly with ice ridge formation on the leading edge upper surface, as the airplane descended from 7,000 feet mean sea level (msl) to 4,000 feet msl in 177 icing conditions; further, this type of ice accretion might have been imperceptible to the pilots. 9. The suggestion in current Federal Aviation Administration publications that “trace” icing is “not hazardous” can mislead pilots and operators about the adverse effects of thin, rough ice accretions. 10. Because the pilots of Comair flight 3272 were operating the airplane with the autopilot engaged during a series of descents, right and left turns, power adjustments, and airspeed reductions, they might not have perceived the airplane’s gradually deteriorating performance. 11.

[ANALYSIS] Although the pilots reacted promptly to the autopilot disengagement and applied control wheel inputs to counter the resultant abrupt left roll, they were not able to regain control of the airplane because of the airplane’s extreme unusual attitude, the highly dynamic nature of the subsequent maneuvers, the presence and effect of ice on the wings, and the low altitude at which the loss of control occurred. The airplane entered an extreme nose-down pitch attitude from which it did not recover. 2.6 FAA Oversight Issues The Safety Board’s investigation of this accident raised concerns about the FAA’s continuing airworthiness oversight of the EMB-120 and the agency’s oversight of icing-related incidents and accidents involving turbopropeller-driven transport airplanes, the adequacy of existing FAA regulatory requirements for the certification of transport-category airplanes for flight into icing conditions (specifically 14 CFR Part 25 appendix C and Section 25.1419), the FAA’s policies for AFM and air carrier operating manual revisions, and the sharing of information related to such revisions between the FAA’s certification and flight standards personnel. 2.6.1 FAA Continuing Airworthiness Oversight Issues The Safety Board notes that, like the ATR-42 and -72, the EMB-120 exhibited a history of icing-related upsets/losses of control before being involved in a related fatal accident. At the time of the Comair accident, six icing-related EMB-120 events had been documented, the first of which occurred in June 1989.187 The Safety Board’s review of these incidents shows that before the Comair accident, the EMB-120 fleet had experienced repeated instances of roll upsets associated with ice accumulations that the pilots either did not observe or did not consider sufficient to prompt activation of the deicing boots. 187 Similarly, before the ATR-72 accident at Roselawn, Indiana, the FAA had been aware of a number of prior ATR upset events. The FAA had concluded that these incidents were essentially pilot-induced stall events; however, further investigation revealed that there were more complex airplane controllability issues involved in the ATR upset events. 164 FAA and Embraer personnel had noted the recurring events, and the FAA presented a summary of the six events at an FAA/industry meeting (attended by Safety Board staff) on November 7, 1995. Further, the FAA and Embraer discussed the events with representatives from Comair and other operators at a meeting on November 15, 1995, and additional discussion took place during the EMB-120 SLD icing tanker tests in December 1995. An FAA engineer reviewed these six incidents in a draft report dated January 26, 1996 (see section 1.18.2.3). The Safety Board has been unable to obtain information about the specific disposition of the draft report within the FAA, although the FAA asserted after the accident that this report did not reflect the official views of the FAA.

[ANALYSIS] Comair’s EMB-120 FSM defined icing conditions as existing “when the OAT [outside air temperature] is +5o C or below and visible moisture in any form is present (such as clouds, rain, snow, sleet, ice crystals, or fog with visibility of one mile or less).” For years, airplane manufacturers have incorporated leading edge deicing boots in the design of airplanes that are to be certificated for operation in icing conditions; the purpose of deicing boots is to shed the ice that accumulates on protected surfaces of the airframe. Over the years, leading edge deicing boots have demonstrated their effectiveness to operators and pilots by keeping the wing and tail leading edges relatively clear of aerodynamically degrading ice accumulations, to the point that operators and pilots have become confident that the airplanes can be flown safely in icing conditions as long as the airplane’s deicing boots are operated (and functioning) properly. However, based on problems with earlier deicing boot designs (which 180 Although Embraer’s nomenclature identifies the propeller ice protection mechanism as a deicing system, it functions as an anti-icing system because it is activated before ice accumulates on the airframe. 149 used larger tubes and lower pressures, resulting in slower inflation/deflation rates), manufacturers, operators, and pilots developed the belief that premature activation of the leading edge deicing boots could (as cautioned in Comair’s EMB-120 FSM) “result in the ice forming the shape of an inflated de-ice boot, making further attempts to deice in flight impossible [ice bridging].” Thus, at the time of the accident, Comair’s (and most other EMB-120 operators’) guidance indicated that pilots should delay activation of the leading edge deicing boots until they observed ¼ inch to ½ inch ice accumulation, despite Embraer’s FAA and Centro Tecnico Aeroespacial of Brazil (CTA) approved EMB-120 Airplane Flight Manual (AFM) revision 43, which indicated that pilots should activate the leading edge deicing boots at the first sign of ice accumulation (see discussion later in this section). The pilots’ activation of the propeller and windshield ice protection systems when the airplane entered the clouds would indicate that they were aware that the airplane was operating in icing conditions. If they had activated the leading edge deicing boots, at least some of the airplane’s degraded performance would have been restored. However, even if the pilots observed any of the thin, rough ice accretion that likely existed before the loss of control, they probably would not have activated the deicing boots because Comair’s guidance to its pilots advised against activating the deicing boots until they observed a thicker ice accumulation.

[ANALYSIS] The Safety Board considers it likely that the pilots would have commented and/or taken action (such as activating the deicing boots and/or extending the flaps) if they had perceived an unsafe condition, either as the result of a significant ice accumulation or an unsafe airspeed assignment for the airplane’s configuration. The Safety Board acknowledges that increasing the airspeed by some increment (Vref + 5 knots according to Comair’s EMB-120 FSM) when ice accretion is observed is a fairly standard adjustment in the aviation industry, and Comair’s FSB 96-04 specified a minimum airspeed of 170 knots for holding in icing conditions. However, ATC had not issued holding instructions to the pilots of Comair flight 3272, nor had ATC indicated that the pilots should expect to receive holding instructions during the approach to DTW. Therefore, the pilots might not have considered the 170-knot minimum airspeed for holding in icing conditions. Additionally, as previously discussed, the pilots might not have recognized that they were operating in icing conditions because it is possible that the accident airplane accreted a thin, rough layer of glaze ice that was imperceptible to the pilots. Because there were no comments recorded by the CVR and because the pilots accepted the 150-knot airspeed assignment without hesitation, comment, or reconfiguration, the Safety Board concludes that the pilots likely did not recognize the need to abide by special restrictions on airspeeds that were established for icing conditions because they did not perceive the significance (or presence) of Comair flight 3272’s ice accumulation. Further, based on the uncertainty regarding minimum airspeeds exhibited by Comair pilots during postaccident interviews, the Safety Board considers it likely that under conditions similar to those encountered by the pilots of Comair flight 3272, other Comair pilots might have accepted the same 150-knot airspeed assignment. Although the Safety Board considers Comair’s airspeed guidance ambiguous and unclear and acknowledges that the flightcrew might not have perceived that the airplane was accumulating ice that affected its flight handling characteristics, the Safety Board notes that the preponderance of the airspeed guidance available to the pilots indicated that EMB-120 operating airspeeds of 160 or 170 knots were standard for operating without flaps extended under any (icing or nonicing) conditions. Though these airspeeds were not established minimum airspeeds, they were the operator’s procedural guidance and the standards to which Comair’s pilots were trained. The Safety Board considers that any pilot deviations from standard procedures during flight operations (although not prohibited and not necessarily unsafe) should be accomplished thoughtfully and with full consideration given to the possible risks involved. In this case, operating at 150 knots provided the pilots with a reduced safety margin above the airplane’s stall speed. The reduction in stall margin was especially critical to the accident flight because the accident airplane had accreted structural ice during its descent, which was having an adverse effect on the airplane’s performance characteristics.

[ANALYSIS] In this case, operating at 150 knots provided the pilots with a reduced safety margin above the airplane’s stall speed. The reduction in stall margin was especially critical to the accident flight because the accident airplane had accreted structural ice during its descent, which was having an adverse effect on the airplane’s performance characteristics. The Safety Board notes that the pilots could have increased the stall margin by extending 15o of flaps and still complied with ATC’s airspeed assignment. Further, there was no safety or operational reason to avoid extending the flaps.185 185 The Safety Board considered the possibility that the flightcrew avoided extending the flaps because of guidance to avoid extended operations in icing conditions with flaps extended. However, as previously discussed, there were numerous indications that the flightcrew was not considering icing as a significant factor in the airplane’s operation at the time. The Safety Board also considered that the pilots might have believed that they had already extended the flaps to 15o at the time that they accepted the 150 knot ATC-assigned airspeed. However, at that time, the airplane was about 20 miles from the destination airport and maintaining an assigned airspeed of 190 knots; thus, the pilots had not received any of the usual (distance and airspeed-related) cues to extend the flaps. 157 The Safety Board considers it critical that pilots take into consideration potential adverse conditions, and make correspondingly conservative decisions where they are warranted. Although the pilots might not have perceived that the airplane was accumulating any ice, their activation of the propeller and windshield heat when the airplane entered icing conditions was an indication that they were aware that they were entering conditions in which ice accumulation was possible. Based on Comair’s guidance for an ILS approach (which Comair uses during pilot training) that associates 170 knots with 15o of flaps on the base leg position, and additional airspeed guidance suggesting airspeeds of 160 to 170 knots for the accident flight’s conditions, and the pilots’ responsibility to make safe, conservative decisions consistent with flight in icing conditions, the Safety Board concludes that whether the pilots perceived ice accumulating on the airplane or not, they should have recognized that operating in icing conditions at the ATCassigned airspeed of 150 knots with flaps retracted could result in an unsafe flight situation; therefore, their acceptance of the 150-knot airspeed assignment in icing conditions without extending flaps contributed to the accident. 2.5.2.3 FAA-related Information Regarding Minimum Airspeeds Because the issue of safe minimum airspeeds is complex and critical to safe flight operations, in May 1997 the Safety Board issued Safety Recommendation A-97-31, which asked the FAA to require air carriers to reflect FAA-approved minimum airspeeds for all flap settings and phases of flight, including flight in icing conditions, in their EMB-120 operating manuals.

[CONCLUSIONS > FINDINGS] The icing certification process has been inadequate because it has not required manufacturers to demonstrate the airplane’s flight handling and stall characteristics under a sufficiently realistic range of adverse ice accretion/flight handling conditions. 28. The work conducted by the Federal Aviation Administration Environmental Icing National Resource Specialist and the Aviation Rulemaking Advisory Committee’s icing-related working groups is of crucial importance to the future safety of icing operations. 29. The potential consequences of operating an airplane in icing conditions without first having thoroughly demonstrated adequate handling/controllability characteristics in those conditions are sufficiently severe that they warrant as thorough a certification test program as possible, including application of revised standards to airplanes currently certificated for flight in icing conditions. 30. The current Federal Aviation Administration policy allowing air carriers to elect not to adopt airplane flight manual operational procedures without clear written justification can result in air carriers using procedures that may not reflect the safest operating practices. 31. At the time of the Comair flight 3272 accident, pertinent flight standards personnel (specifically, the principal operations inspector assigned to Comair) lacked information critical to the continued safe operation of the EMB-120 fleet and would have been unable to evaluate the need to 180 incorporate airplane flight manual revision 43 or any alternatives proposed by air carriers. 32. The Federal Aviation Administration’s current EMB-120 flight data recorder system inspection procedure is inadequate because it allows existing flight control sensor anomalies to go undetected, and thus uncorrected. 33. The failure of pilots who encounter in-flight icing to report the information to the appropriate facility denies other pilots operating in the area the access to valuable and timely information that could prevent an accident. 34. The Federal Aviation Administration air traffic control system has not established adequate procedures for the dissemination of icing-related pilot reports received in the airport terminal environment; these reports should be incorporated into automatic terminal information service broadcasts so that all arriving and departing pilots can become aware of icing conditions in the area. 181 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the FAA’s failure to establish adequate aircraft certification standards for flight in icing conditions, the FAA’s failure to ensure that a Centro Tecnico Aeroespacial/FAAapproved procedure for the accident airplane’s deice system operation was implemented by U.S.- based air carriers, and the FAA’s failure to require the establishment of adequate minimum airspeeds for icing conditions, which led to the loss of control when the airplane accumulated a thin, rough accretion of ice on its lifting surfaces. Contributing to the accident were the flightcrew’s decision to operate in icing conditions near the lower margin of the operating airspeed envelope (with flaps retracted), and Comair’s failure to establish and adequately disseminate unambiguous minimum airspeed values for flap configurations and for flight in icing conditions. 182

[ANALYSIS] Consequently, the Safety Board reviewed the adequacy of the current FAA requirements for the certification of airplanes for flight in icing conditions. For an airplane to be certificated for flight in icing conditions, the FAA requires the manufacturer to demonstrate a limited number of test data points within the Part 25 appendix C envelope. The FAA’s icing certification requirements are based on fully functioning and operating anti-icing and deicing systems. Although there is no requirement for manufacturers to consider the effects of delayed activation of ice protection systems, intercycle or residual ice accumulations, or other variables that might result in significant aerodynamic effects, Embraer exceeded the minimum FAA requirements when Embraer tested the EMB-120 with ¾-inch (U.S.) and 1-inch (Canada) ice accretions/shapes during initial icing certification.189 Certification records indicate that the EMB-120 successfully exhibited satisfactory flight handling characteristics with 3-inch ram’s horn ice shapes installed on unprotected surfaces. Further, during the SLD icing controllability tests, the FAA tested the EMB-120 with quarter-round artificial ice shapes as large as 1 inch located at the aft edge of the farthest aft inflatable deicing boot segment (to represent ice accumulated in icing conditions that fall outside the Part 25 appendix C envelope). The airplane exhibited full lateral controllability and satisfactory stall warning characteristics in this condition.190 However, Embraer had not demonstrated (nor was the company required by the certification authorities to demonstrate) the EMB-120’s performance in other ice configurations that would result from weather conditions within the Part 25 appendix C LWC and droplet size 189 For U.S. (FAA) icing certification, the EMB-120 was tested with ¼ inch, ½ inch, and ¾ inch of natural ice on protected surfaces, up to 4 inches of natural ice accumulation on unprotected airfoil surfaces, and 3inch ram’s horn artificial ice shapes on unprotected surfaces; except for the ¾-inch natural ice on protected surfaces, these conditions could be encountered while operating in icing conditions in accordance with procedures outlined in the EMB-120 AFM. However, for Canadian icing certification, the EMB-120 was tested with artificial ice shapes representing conditions considered to be outside normal operation with deicing boots activated (1-inch ram’s horn ice shapes on protected surfaces). 190 Although some control wheel force exceedences were observed, tanker tests identified more realistic ice shapes; during subsequent tests with the realistic ice shapes, no excessive control wheel forces or other anomalies were noted. 167 envelope, including realistic ice shapes (or natural ice) representing a thin layer of sandpapertype ice with a small ice ridge (as may have been experienced by Comair flight 3272). As discussed in section 2.4, postaccident icing and wind tunnel information indicated that with a small ice ridge along that thin rough surface, the aerodynamic effect on handling and stall margin/stall warning (reduced stall AOA and rapid decrease in lift) can be worse than any of the ice shapes that the FAA required for icing certification.

[ANALYSIS] If they had activated the leading edge deicing boots, at least some of the airplane’s degraded performance would have been restored. However, even if the pilots observed any of the thin, rough ice accretion that likely existed before the loss of control, they probably would not have activated the deicing boots because Comair’s guidance to its pilots advised against activating the deicing boots until they observed a thicker ice accumulation. Therefore, based on CVR information and on the steady degradation of airplane performance that was clearly uninterrupted by leading edge deicing boot activation, the Safety Board concludes that, consistent with Comair’s procedures regarding ice protection systems, the pilots did not activate the leading edge deicing boots during their descent and approach to the Detroit area, likely because they did not perceive that the airplane was accreting significant (if any) structural ice. During the postaccident (November 1997) Airplane Deicing Boot Ice Bridging Workshop, information regarding recent icing tunnel and flight test research into the ice bridging phenomenon was disseminated and discussed among industry personnel (see section 1.18.4.2). The recent research revealed that modern turbine-powered airplanes, with their high-pressure, segmented pneumatic deicing boots, are not at risk for ice bridging.181 However, in April 1996 when Embraer issued (FAA- and CTA-approved) revision 43 to the EMB-120 AFM, the procedure it recommended—activation of the leading edge deicing boots at the first sign of ice accretion—was not consistent with traditional industry concerns about ice bridging. According to the FAA’s EMB-120 Aircraft Certification Program Manager, when the EMB-120 AFM revision was proposed by Embraer in late 1995, the deicing boot procedural change was very controversial and generated numerous discussions among FAA and industry personnel. The FAA’s EMB-120 Aircraft Certification Program Manager stated that the aircraft evaluation group (AEG) personnel involved in the discussions about the six EMB-120 icing-related events, the EMB-120 in-flight icing tanker tests, and the deicing boot procedural change were initially resistant to the deicing boot procedural change because of the perceived potential for ice bridging. 181 It is important to note that ice bridging may still be a potential hazard for airplanes with older technology deicing boots that have slower inflation/deflation rates. 150 The Safety Board notes that during the winter of 1995/1996, senior Comair personnel (and representatives from other EMB-120 operators) were involved in numerous meetings and discussions regarding the six preaccident icing-related events and that they subsequently received Embraer’s Operational Bulletin (OB) 120-002/96 and revision 43 to the EMB-120 AFM, with its controversial deicing boot procedural change.

[ANALYSIS] The studies showed that this effect was most pronounced when a leading edge ridge of ice was present. • differences in local airflow over left and right wings because of propeller thrust—both propellers rotating in a clockwise direction results in asymmetrical thrust and a left yaw tendency. • the airplane leveling off at its assigned altitude and slowing to its assigned airspeed resulted in an increased AOA and movement of the stagnation point on the leading edge of both wings, which might have exacerbated the effects listed above. • left yaw resulting from the asymmetrical engine power application that occurred less than 3 seconds before the autopilot automatically disengaged. Flight simulations that varied the timing and symmetry of the engine power application indicated that when power was applied earlier and/or more symmetrically, the simulator’s bank did not exceed the autopilot’s command limit, the autopilot did not abruptly disengage, and the upset did not occur. However, flight simulations indicated that the asymmetrical engine power application observed in the accident FDR data would not have resulted in an upset if the aerodynamic degradation from ice was not present. (See sections 1.16.1.2 and 2.2.) • the effects of the autopilot disengagement—when the autopilot (which had been commanding RWD aileron to resist the left roll) automatically disengaged, the sudden absence of resistance resulted in a significant increase in the left roll and roll rate. (See section 2.5.4.) The Safety Board also considered the possibility that the absence of conductive edge sealer179 on the left wing leading edge segments (see section 1.12.2) might have contributed to an asymmetrical ice accretion that would have increased the left roll tendency. Postaccident examination revealed that the leading edge segments that did not have the conductive edge sealer applied were as smooth to the touch as any other part of the deicing boot/leading edge; further, there was no roughness, cracking, splitting, or delamination observed in the area where the conductive edge sealer should have been. Safety Board staff asked the scientists at NASA-Lewis 178 NASA’s two-dimensional studies indicated that such a reversal is possible at small aileron deflections and higher AOAs; FDR data indicated that during the last minute before the upset, the aileron deflection and AOA was frequently changing. 179 It was not possible to determine when and/or how the conductive edge sealer came to be missing; whether it eroded or wore off during flight, during the accident sequence, as a result of application of ICEX, or whether the conductive edge sealer was not applied properly during deicing boot installation.

[ANALYSIS] Thus, the Safety Board believes that the FAA should require the manufacturers and operators of all airplanes that are certificated to operate in icing conditions to install stall warning/protection systems that provide 186 FAA and NASA wind/icing tunnel data indicate that the NACA 23012 airfoil with a thin layer of rough ice on the leading edge with a small ice ridge can stall at angles of attack as low as 5o or 6o. 160 a cockpit warning (aural warning and/or stick shaker) before the onset of stall when the airplane is operating in icing conditions. 2.5.4 Operation of the Autopilot The Safety Board was unable to positively determine whether the autopilot was operating properly based on physical evidence (impact damage precluded functional tests). However, based on FDR data and a review of the autopilot design characteristics, the Safety Board concludes that the accident airplane’s autopilot was capable of normal operation and appeared to be operating normally during the last minutes of the accident flight, and the autopilot disconnect and warning systems operated in a manner consistent with their design logic. The Safety Board evaluated the flightcrew’s use of the autopilot as it affected the cues presented to the pilots about the impending loss of control and the behavior of the ailerons as the loss of control developed. The autopilot’s actions during the last seconds before it disengaged provided some visual cues that could have warned the pilots of the airplane’s performance degradation. For example, during the 15 seconds before the autopilot disengaged, it moved the control wheel to command the ailerons to move in a RWD direction, while the flight instruments (EADIs) and the pilots’ heading selection indicated that the airplane was in a left bank. Although it would have been possible for the pilots to observe this and deduce that an anomalous flight condition existed, these visual cues began very gradually and were subtle and short lived. The control wheel did not move more than 10o, and the roll angle did not exceed 30o (only slightly greater than the normal autopilot bank limit for the selected left turn), until about 8 seconds before the upset. The deviations from the desired airplane attitude were becoming noticeable about the time that the pilots were increasing engine power to maintain 150 knots and continued as the captain directed the first officer’s attention to the airplane’s airspeed (about 5 seconds before the upset). Given this distraction, it is likely that the subtle visual cues that were available were not adequate to prompt the pilots to take the direct and aggressive action that would have been necessary to avoid the upset.

[ANALYSIS] When the captain drew the first officer’s attention to the low airspeed indication at 1554:20.8, the airplane’s airspeed had decreased to 147 knots. During the next 2 seconds, the pilots more aggressively increased the engine power, and a significant torque split occurred; the torque values peaked at 108 percent on the left engine and 138 percent on the right engine. The Safety Board considered several possible reasons for the significant torque split, including uneven throttle movement by the pilots, ice ingestion by the left engine, a misrigged engine, or an improper engine trim adjustment on the newly installed right engine; however, it was not possible to positively determine the cause of the torque split. Postaccident simulations indicated that this torque split had a significant yaw-producing effect at a critical time in the upset event, exacerbating the airplane’s excessive left roll tendency (see section 2.4.1). The airplane’s airspeed decreased further to 146 knots, the left roll angle increased beyond the autopilot’s 45o limit, and (at 1554:24.1) the autopilot disconnect warning began to sound. One second later, the stick shaker activated. The sudden disengagement of the autopilot (at 1554:24.125) greatly accelerated the left rolling moment that had been developing, suddenly putting the airplane in an unusual attitude. Although the pilots were likely surprised by the upset event, interpretation of the FDR data indicated that the pilots responded with control wheel inputs to counter the left roll within 1 second of the autopilot disengagement and continued to apply control inputs in an apparent attempt to regain control of the airplane until the FDR recording ceased. 2.3 Meteorological Factors Although Comair flight 3272 was operating in winter weather conditions throughout its flight from the Cincinnati area to Detroit, CVR and weather information indicated that the airplane was operating above the cloud tops at its cruise altitude of 21,000 feet msl. Further, the temperatures at the altitudes flown during the en route phase of the flight were too cold to be conducive to airframe ice accretion, and examination of the FDR data did not reflect degraded airplane performance until later in the airplane’s descent (see section 1.16.1.1). Therefore, the Safety Board concludes that the airplane was aerodynamically clean, with no effective ice accreted, when it began its descent to the Detroit area. A study conducted by the National Center for Atmospheric Research (NCAR) indicated that there was strong evidence for the existence of icing conditions in the clouds along the accident airplane’s descent path below 11,000 feet msl.

[ANALYSIS] For example, the American Eagle flight 4184 accident investigation revealed that SLD conditions can cause ice accretions that are more aerodynamically detrimental than those accretions that fall within the Part 25, Appendix C envelope. As a result, the Safety Board issued Safety Recommendation A-96-54, which asked the FAA to do the following: Revise the icing criteria published in 14…CFR Parts 23 and 25, in light of both recent research into aircraft ice accretion under varying conditions of liquid water content, drop size distribution, and temperature, and recent developments in both the design and use of aircraft. Also, expand the Appendix C icing certification envelope to include freezing drizzle/freezing rain and mixed water/ice crystal conditions, as necessary. 61 AC 23.1419-2C stated that the effect on pilot workload of continuously cycling the deice boots should be evaluated. Analysis 35 Aircraft Accident Report Further, icing tunnel tests conducted as part of the Comair flight 3272 accident investigation indicated that the effects of ice accretion on airplane performance could vary widely depending on the size, distribution, and type of ice accumulated on the airplane’s surfaces. However, the Board learned that manufacturers are not required to demonstrate an airplane’s flight handling characteristics or stall margins using thin, rough ice that can accrete on protected surfaces before the activation of the deice boot system or between activation cycles. As a result of its findings, the Board issued Safety Recommendation A-98-92, which asked the FAA (in cooperation with NASA and other interested aviation organizations) to do the following: [C]onduct additional research to identify realistic ice accumulations and … determine the effects and criticality of such ice accumulations … the information developed … should be incorporated into aircraft certification requirements. The Safety Board also issued Safety Recommendation A-98-100, which asked the FAA to review the icing certification of all turbopropeller-driven airplanes currently certificated for operation in icing conditions, perform additional testing, and take action as required to ensure that these airplanes fulfill the requirements of the revised icing certification standards asked for in Safety Recommendation A-98-92. The FAA indicated in a March 6, 2006, response to Safety Recommendation A-96-54 that the ARAC IPHWG is continuing to develop a revision to Part 25 to require a demonstration that an airplane can safely operate in SLD conditions for an unrestricted time or can detect SLD and safely exit icing conditions. However, the FAA has still not received the recommendations from the IPHWG, prepared regulatory analyses, issued the NPRM, analyzed comments, or completed the many other tasks involved in issuing new regulations.

[ANALYSIS] During the approach, the flight crew of the sister ship, which was following the accident flight, cycled the deice boots numerous times and maintained a high airspeed and, subsequently, landed safely, indicating the importance of taking these actions to counteract the hazardous effects of icing. This analysis discusses the accident sequence, including the flight crew’s performance. This analysis also discusses inadequate training on operations in icing conditions, inadequate deice boot system operational guidance, the need for automatic deice boot systems, inadequate icing flight test certification requirements, and inadequate stall warning margins in icing conditions. 2.2 Accident Sequence 2.2.1 Descent Into Icing Conditions Surface observations and radar data indicated that freezing drizzle conditions existed in the PUB area around the time of the accident and that temperatures were below freezing from the surface to 30,000 feet. Several PIREPs and NWS products transmitted around the time of the accident, including winter weather advisories that warned of freezing drizzle, confirmed the presence of icing conditions in the PUB area. CVR Analysis 28 Aircraft Accident Report evidence indicates that, starting about 0851, the flight crew began taking actions to minimize the icing’s hazardous effects, such as activating the airplane’s engine and windshield anti-ice systems. About 4 minutes later, the first officer stated that he saw a little bit of “grayish ice,” which is indicative of ice that has a rough surface, building up on the wing leading edges. He then stated that there was “a real thin line back there.” An analysis of the CVR and meteorological information indicated that mixed icing conditions existed from about 21,000 to 14,000 feet. Radar data and CVR information indicated that the airplane was in this icing layer for about 5 1/2 minutes. At 0858:20, as the airplane was descending through about 18,000 feet, the first officer suggested to the captain that he might want to cycle the deice boots.54 After cycling the deice boots, the captain indicated that the deice boots might have shed a little of the ice but that some ice remained on the wing, indicating the presence of residual ice. 2.2.2 Approach to Landing In accordance with the SimuFlite Cessna Citation V Technical Manual and the Cessna Model 560 Citation V AFM, pilots were trained that, when any residual ice is present or can be expected during approach and landing, Vref must be increased by 8 knots. The manuals also contained both a caution and a warning indicating that stall speeds increased during operations in icing conditions, and that, therefore, Vref must be increased. At 0859:29, the CVR recorded the first officer state that the Vref was 96 knots. In the case of this flight, the Vref should have been increased from 96 to 104 knots because of the icing conditions. The CVR did not record either pilot mention increasing the airspeed at any point during the approach.

[CONCLUSIONS > FINDINGS] Therefore, the Safety Board believes that the FAA should require modification of the Cessna 560 airplane’s stall warning system to provide a stall warning margin that takes into account the size, type, and distribution of ice, including thin, rough ice on or aft of the protected surfaces. 38 Aircraft Accident Report 3. Conclusions 3.1 Findings 1. The captain and first officer were properly certificated and qualified under Federal regulations. No evidence indicated any preexisting medical or physical condition that might have adversely affected the flight crew’s performance during the accident flight. A review of the pilots’ 72-hour histories revealed that they slept well in the days leading up to the accident flight and went to bed early in preparation for an early departure. No evidence was found that fatigue degraded the performance of either pilot on the day of the accident. 2. The weight and balance of the airplane were within landing limits. 3. The recovered components showed no evidence of preexisting powerplant, structural, or system failures. 4. The Pueblo Memorial Airport local controller did not provide the accident flight crew or the Denver Flight Service Station with the pilot report reporting light to moderate icing; however, this was not a factor in the accident because cockpit voice recorder information indicated that the flight crew was aware of the icing conditions. 5. During the approach, the flight crew of the sister ship, which was following the accident flight, cycled the deice boots numerous times and maintained a high airspeed and, subsequently, landed safely, indicating the importance of taking these actions to counteract the hazardous effects of icing. 6. The flight crew did not increase the landing reference airspeed while operating in icing conditions, contrary to company procedures and manufacturer guidance. 7. The airplane encountered supercooled large droplet (SLD) conditions, which are most conducive to the formation of thin, rough ice on or aft of the protected surfaces, during about the last 4 1/2 minutes of the flight. Further, the airplane had residual ice on the wings after the deice boots were activated earlier in the flight, and this ice would have affected the overall thickness, roughness, and distribution of the SLD ice accumulation. 8. The flight crew did not activate the deice boots when configuring for the approach and landing, which was contrary to company procedures and manufacturer guidance. 9. The flight crew failed to maintain adequate airspeed during the final approach in icing conditions, which led to an aerodynamic stall from which they did not recover. Conclusions 39 Aircraft Accident Report 10. Pilots could benefit from the reinforcement during training of the Cessna Model 560 Citation V Airplane Flight Manual requirements to increase the airspeed and operate the deice boots during approaches when ice is present on the wings. 11. The briefing conducted late in the approach was a distraction that impeded the flight crew’s ability to monitor and maintain airspeed and manage the deice system. 12. All operators would benefit from an increased focus on providing monitoring skills in their training programs, including those operating under 14 Code of Federal Regulations Parts 121 and 135, as would pilots completing Federal Aviation Administration-approved training programs for Part 91 operations. 13.

[ANALYSIS] Analysis 33 Aircraft Accident Report flight 3272 investigation, the Safety Board learned that many manufacturers and operators had similar deice boot operational guidance and concerns about ice bridging. However, AC 25.1419-1A states that, although ice may not be completely shed by one cycle of the boots, the residual ice will usually be removed by subsequent cycles and does not act as a foundation for a bridge of ice to form. Further, information gathered at a 1997 Airplane Deice Boot Bridging Workshop, subsequent icing tunnel tests, and flight tests conducted as part of the Comair investigation revealed that ice bridging did not occur on modern airplanes, which are equipped with deice boots that quickly inflate and deflate. The icing tunnel tests also revealed that thin (1/4 inch or less), rough ice accumulations on the wing leading edge deice boot surfaces could be, depending on distribution, as aerodynamically detrimental to an airplane’s performance as larger ice accumulations. A search of the Safety Board accident database revealed no accidents related to ice bridging. Conversely, the Board has investigated many icing accidents in which the airplane stalled prematurely and the stall warning system did not activate before the stall because of ice accumulation on the wing leading edges. This accident, previous accident investigations, Safety Board accident data, and existing icing information clearly show that delaying the activation of the deice boots can create unsafe operations. The Safety Board concludes that ice bridging does not occur on modern airplanes; therefore, it is not a reason for pilots to delay activation of the deice boots. As a result of its findings during the Comair flight 3272 investigation, the Safety Board issued Safety Recommendation A-98-91, which recommended that the FAA do the following: Require manufacturers and operators of modern turbopropeller-driven airplanes in which ice bridging is not a concern to review and revise the guidance contained in their manuals and training programs to emphasize that leading edge deicing boots should be activated as soon as the airplane enters icing conditions. In May 2002, the FAA issued an icing test report that recommended an “early and often” approach to deice boot usage to limit the size of residual and intercycle ice accretions. Further, in January 2003, an ARAC IPHWG recommended revisions to Parts 25 and 121 to require that deice systems be activated as soon as an airplane enters icing conditions. However, since that time, the FAA has taken no action to issue a final rule adopting the regulatory changes proposed by the ARAC IPHWG. Although the accident airplane most likely accumulated less than 1/4-inch-thick ice while operating in the lower cloud layer, the pilots’ failure to activate the deice boots during the approach led to the continued accumulation of thin, rough ice on the protected surfaces, which can severely degrade an airplane’s performance.

[ANALYSIS] The 1996 ice shapes tests on the Cessna 560 were also inadequate because, although tests were conducted with ice shapes on the protected surfaces, tests were not conducted using thin, rough ice. Therefore, additional ice sizes, distribution patterns, and types need to be considered during flight testing to more adequately gauge an airplane’s performance in icing conditions. The Safety Board concludes that existing flight test certification requirements for flight into icing conditions do not test the effects of thin, rough ice on or aft of an airplane’s protected surfaces, which can cause severe aerodynamic penalties. The circumstances of this accident clearly show that the actions requested in Safety Recommendations A-96-54 and A-98-92 are needed to improve the safety of all airplanes operating in icing conditions. Therefore, the Safety Board reiterates Safety Recommendations A-96-54 and A-98-92. As noted, Safety Recommendation A-98-100 only addressed turbopropeller-driven airplanes. The circumstances of this accident clearly demonstrate that deice boot-equipped turbojet airplanes also require additional testing in an expanded Appendix C icing certification envelope, which would include thin, rough ice accumulations and intercycle and residual ice. Therefore, the Safety Board believes that the FAA should, when the revised icing certification standards and criteria are complete, review the icing certification of all pneumatic deice boot-equipped airplanes that are currently certificated for operation in icing conditions and perform additional testing and take action as required to ensure that these airplanes fulfill the requirements of the revised icing certification standards. The new recommendation (A-07-16) will supersede Safety Recommendation A-98-100 and will be classified “Open—Unacceptable Response.” 62 2.5 Inadequate Stall Warning Margins in Icing Conditions Stall warning systems are intended to provide flight crews with adequate warning of an impending stall to give them enough time to take necessary action to prevent a stall. 62 Safety Recommendation A-98-100 is on the Safety Board’s List of Most Wanted Transportation Safety Improvements. Accordingly, Safety Recommendation A-07-16 will automatically be placed on the Most Wanted List. Analysis 37 Aircraft Accident Report The CVR sound spectrum study indicated that the accident airplane’s stall warning did not activate until after the stall. The Pueblo accident is not the first accident in which a stall has occurred before the stall warning activated. For example, the Safety Board determined during the Comair flight 3272 accident investigation that the airplane departed controlled flight before the stall warning activated and that stall warning systems “often do not provide adequate warning when the airplane is operating in icing conditions.” As a result of the Comair investigation, the Safety Board issued Safety Recommendation A-98-96, which recommended that the FAA require manufacturers and operators of all airplanes certificated to operate in icing conditions to install stall warning systems that provide a cockpit warning before the onset of a stall when the airplane is operating in icing conditions.

[ANALYSIS] At 0859:29, the CVR recorded the first officer state that the Vref was 96 knots. In the case of this flight, the Vref should have been increased from 96 to 104 knots because of the icing conditions. The CVR did not record either pilot mention increasing the airspeed at any point during the approach. Therefore, the Safety Board concludes that the flight crew did not increase the Vref while operating in icing conditions, contrary to company procedures and manufacturer guidance. 2.2.3 Final Approach At 0908:25, while at an altitude of about 9,400 feet, the first officer reported that the flight was in IMC, and, about 1 minute later, while at an altitude of about 7,400 feet, he reported that clear ice had accumulated on the airplane’s wing. CVR and meteorological information indicated that the airplane likely encountered SLD conditions from 9,400 to 6,100 feet (the calculated altitude at the time of the upset) and that the airplane was likely in these conditions for about 4 1/2 minutes. During this time, about 1 to 4 mm (0.039 to 0.156 inch) of additional ice could have accumulated on the wing leading edges. The Safety Board concludes that the airplane encountered SLD conditions, which are most conducive to the formation of thin, rough ice on or aft of the protected surfaces, during about the last 4 1/2 minutes of the flight. The Safety Board further concludes that the airplane had residual ice on the wings after the deice boots were activated earlier in the 54 As noted, the pilots had been trained to wait until 1/4- to 1/2-inch-thick ice accumulation was visible on the wing leading edges before activating the deice boots. Analysis 29 Aircraft Accident Report flight and that this ice would have affected the overall thickness, roughness, and distribution of the SLD ice accumulation. According to the airplane performance study, about 0910, the airplane started its final descent from 7,000 feet at an airspeed of about 155 knots. By about 0911:35, the airspeed had started to decrease. CVR evidence indicated that the landing gear was extended at 0911:10, followed by extension of the speedbrakes and selection of full flaps. At 0912:04, the first officer stated, “and you are plus twenty five,” to which the captain replied, “slowing.” On the basis of a Vref of 96 knots, the airspeed would have been about 121 knots at the time of the first officer’s statement. At 0912:37, when the airplane was at an altitude of about 6,100 feet, the first officer told the captain that he might want to run the deice boots and that they had the Vref.

[CONCLUSIONS] The briefing conducted late in the approach was a distraction that impeded the flight crew’s ability to monitor and maintain airspeed and manage the deice system. 12. All operators would benefit from an increased focus on providing monitoring skills in their training programs, including those operating under 14 Code of Federal Regulations Parts 121 and 135, as would pilots completing Federal Aviation Administration-approved training programs for Part 91 operations. 13. Ice bridging does not occur on modern airplanes; therefore, it is not a reason for pilots to delay activation of the deice boots. 14. Activating the deice boots as soon as an airplane enters icing conditions provides the greatest safety measure. 15. Manual operation of the deice boot system increases pilot workload, which can result in distraction during critical phases of flight, such as approach and landing. 16. Existing flight test certification requirements for flight into icing conditions do not test the effects of thin, rough ice on or aft of an airplane’s protected surfaces, which can cause severe aerodynamic penalties. 17. The Cessna 560 airplane’s stall warning system did not provide a stall warning before the upset. 18. The Cessna 560 airplane’s stall warning system does not provide a warning in all icing conditions, including those conditions in which thin, rough ice can accumulate on the protected surfaces. 3.2 Probable Cause The National Transportation Safety Board determines that the probable cause of this accident was the flight crew’s failure to effectively monitor and maintain airspeed and comply with procedures for deice boot activation on the approach, which caused an aerodynamic stall from which they did not recover. Contributing to the accident was the Federal Aviation Administration’s failure to establish adequate certification requirements for flight into icing conditions, which led to the inadequate stall warning margin provided by the airplane’s stall warning system. 40 Aircraft Accident Report 4. Safety Recommendations 4.1 New Safety Recommendations As a result of its investigation of the February 16, 2005, accident involving a Cessna Citation 560, the National Transportation Safety Board makes the following recommendations to the Federal Aviation Administration: Require that operational training in the Cessna 560 airplane emphasize the airplane flight manual requirements that pilots increase the airspeed and operate the deice boots during approaches when ice is present on the wings. (A-07-12) Require that all pilot training programs be modified to contain modules that teach and emphasize monitoring skills and workload management and include opportunities to practice and demonstrate proficiency in these areas. (A-07-13) Require manufacturers and operators of pneumatic deice boot-equipped airplanes to revise the guidance contained in their manuals and training programs to emphasize that leading edge deice boots should be activated as soon as the airplane enters icing conditions. (A-07-14) (This safety recommendation supersedes Safety Recommendation A-98-9163 and is classified “Open—Unacceptable Response.”) Require that all pneumatic deice boot-equipped airplanes certified to fly in known icing conditions have a mode incorporated in the deice boot system that will automatically continue to cycle the deice boots once the system has been activated. (A-07-15) When the revised icing certification standards (recommended in Safety Recommendations A-96-54 and A-98-92) and criteria are complete, review the icing certification of pneumatic deice boot-equipped airplanes that are currently certificated for operation in icing conditions and perform additional testing and take action as required to ensure that these airplanes fulfill the requirements of the revised icing certification standards. (A-07-16) (This safety recommendation supersedes Safety Recommendation A-98-10064 and is classified “Open—Unacceptable Response.”) 63 For more information about this recommendation, see sections 1.18.2.1 and 2.3. 64 For more information about this recommendation, see sections 1.18.2.2 and 2.4.

[ANALYSIS] However, the FAA has still not received the recommendations from the IPHWG, prepared regulatory analyses, issued the NPRM, analyzed comments, or completed the many other tasks involved in issuing new regulations. The FAA indicated in an October 26, 2005, response to Safety Recommendation A-98-92 that it had completed and would shortly issue a draft revision to AC 20-73, which included the certification guidance on determining critical ice shapes, descriptions of intercycle and residual ice accretions, and the aerodynamic penalties associated with these ice shapes. Although the FAA issued AC 20-73A on August 16, 2006, it has still not provided the Safety Board with information regarding any new research conducted in response to this recommendation. Regarding Safety Recommendation A-98-100, the FAA issued an NPRM in November 2005, which proposed to expand 14 CFR Part 25 to include specific certification requirements for airplane performance or handling qualities for flight in icing conditions and to specify the ice accumulations that must be considered for each phase of flight. Further, the FAA proposed changes to AC 25-1X, which intended to provide guidance for implementing the regulations proposed in the NPRM. In May 2006, the Safety Board expressed concern that, although it agreed with the proposed regulatory changes, the FAA had not applied the new standards to all in-service turbopropeller-driven aircraft. The FAA further indicated that no airplanes have an unsafe condition in icing environments despite a number of accidents in the 1990s that involved Analysis 36 Aircraft Accident Report airplanes that had passed the certification standards. The Board stated that, to meet the intent of Safety Recommendation A-98-100, the FAA would need to formally evaluate (perhaps by conducting flight tests) all in-service turbopropeller-driven aircraft to ensure that these aircraft comply with all current icing certification criteria for new aircraft. The Board asked the FAA to provide a list of the aircraft that it had formally evaluated and a summary of the findings and resultant actions. To date, this information has not been received. The circumstances of the Comair flight 3272, American Eagle 4184, and Pueblo accidents and the icing tunnel test data show that the ice shapes used during initial certification flight tests were not adequate because the tests did not account for thin, rough ice on the wing. The 1996 ice shapes tests on the Cessna 560 were also inadequate because, although tests were conducted with ice shapes on the protected surfaces, tests were not conducted using thin, rough ice. Therefore, additional ice sizes, distribution patterns, and types need to be considered during flight testing to more adequately gauge an airplane’s performance in icing conditions.

[FINDINGS] The captain did not use a USAir-approved V, speed. 75 The first officer called Vp L1 knots early, and the captain rotated about 5 knots early. His rotation rate was about 2.5 degrees per second. The airplane accelerated normally during the takeoff roll. After liftoff and before transitioning to the initial climb, the wing stalled before the stall warning system activated. Lateral instability was caused by an irregular stall progression across the wing that led to an abrupt left roll and wing tip strike that further reduced the ability to climb. The airplane experienced a wing lift deficiency because of ice contamination. The initiation of rotation for takeoff at a speed about 5 knots below the prescribed speed resulted in a higher peak AOA at liftoff and, with the wing contamination, eliminated any AOA stall margin that might have existed with a normal rotation. According to wind tunnel studies conducted by the manufacturer, a wing upper surface roughness consisting of particles only 1-2 mm diameter (0.04-0.08 inch), at a density of about one particle per square centimeter, can cause lift losses on the F-28 wing of about 22 and 33 percent, in ground effect and free air, respectively. The first officer observed the wing from the cockpit and stated that he checked the black strip for ice accumulation. The black strip was intended to aid in detection of in-flight leading edge ice and, because of its location on the leading edge, is not effective for detecting upper surface ice. At night, flightcrews cannot visually detect minute amounts of ice on the part of the wing that is visible from the cockpit windows. This part of the wing is 30 to 40 feet from the cockpit. Flightcrews also may not be able to detect such contamination from the cabin windows. 76 Runway 13/3! was not significantly contaminated, and runway conditions were properly reported at LaGuardia on the night of the accident. Accident history shows that nonstatted, turbojet, transportcategory airplanes have been involved in a disproportionate number of takeoff accidents where undetected upper wing ice contamination has been cited as the probable cause or sole contributing factor. No specific injury pattern could be identified in the cabin to explain why some passengers survived the accident and others did not. Passengers who sustained minor injuries and injuries that were not life threatening most likely drowned as a result of confuston, disorientation or entrapment or a combination of these factors. At the time of the accident, procedures for opening emergency door exits were inaccurately and incompletely displayed on USAir's F-28 passenger safety briefing cards, but they did not contribute to the fatalities in the accident. The locations of the dike, pump house, and ILS localizer ground plane antenna were within current FAA guidelines; however, the locations did not meet [CAO Annex 14 criteria.

[ANALYSIS] As with many other domestic air carriers, USAir had not equipped any of its facilities to dispense the Type II fluids to provide extended anti- ice protection to its aircraft. The Safety Board believes that USAir's procedures met airline standards and were consistent with most of the industry. The groundcrews believed that visual inspections were sufficient for determining the presence of airplane surface contamination. Groundcrews interviewed by Safety Board investigators were very conscientious; however, they, like the individuals that trained them, were unaware of the need for tactile inspections under certain conditions. Nonetheless, the Safety Board believes that flight 405 was probably free of frozen contaminants when it left the gate. 2.6.2 Guidance to Flightcrews USAir flightcrews received materials and training concerning winter operations consistent with, and in some cases, exceeding industry standards. The initial F-28 ground school emphasized the critical nature of the F-28 hard wing. The bimonthly publication Flightcrew View provided reference material on many subjects for the flightcrews and was part of their recurrent training program. The September-October 1991 issue contained information on winterization procedures, including AC 20-117. The USAir pilots were also given an examination that included questions about the effects of frost and ice and pilot responsibilities during their recurrent training. Additionally, an excellent perspective of the contamination problem was offered in a memorandum written by an Empire Airlines?> captain in 1984, and issued by the USAir F-28 Flight Manager in November 1991. The captain points out: Contamination - Frost accumulations of as little as t/16 of an inch, like medium to course grit sardpaper, on the wing leading edge can increase stall speeds by 30 percent (right in the vicinity of V,, Vp). Uneven contamination across the leading edge will result in wing drop or roll off as the stall develops across the wing....Ice or frost accumulations «an appear on leading edges during taxi out or takeoff roll - a de-icing beforehand even on a clean wing may prevent such accretion. The captain further wrote that leading edge lift devices recover lift loss due to light ice accumulations. He cautioned that pilots must not get a false sense of security when preceding B-727s successfully take off, especially when their own airplanes are not equipped with leading edge lift devices. SS - TRS 2SEmpire Aitlines vas a regional commuter airline based out of Utica, New York. operating F-28s. It merged with Piedmont Aislires, which subsequently merged with US Air Inc. 58 Finally, the captain pointed out: When wing contamination is suspected despite earlier preventative measures, rotation rates must not be excessive and takeoff speeds may be increased up to 10 knots (increased Vp speeds of up to 10 knots have limited effect on aircraft perfonmance profiles - but speeds in excess of 10 knots adversely effect performance rates). Available field length must be accounted for in the decision to rotate slower than 3° per second and to target higher takeof! speeds.

[ANALYSIS] According to the Fokker simulation data, at this speed, the airplane should have been able to sustain a load factor of 1.5 G at the stick shaker threshold AOA which would still have provided about a 3-degree AOA stall margin. The single “beep” of the aural stall waming immediately after stick shaker activation indicates that the airplane momentarily attained an even higher AOA, between 12.5 and 15 degrees. However, the signal was not continuous, and for 5 seconds the airplane was apparently at an AOA less than that at which lift, with a clean wing, normally begins to decay and drag increases rapidly. That the airplane was unable to attain this nonnal flight perfonnance is considered by the Safety Board to be conclusive evidence that the normal aerodynamic lift capability characteristic of the wing was significantly degraded by an accumulation of frozen contaminant. 2.4 Deicing Fluid Holdover Time and Ice Accumulation The Safety Board found that the airplane had been properly cleared of ice and snow during the two deicing procedures at the gate. However, approximately 35 minutes elapsed between the second time that the airplane was deiced and the initiation of takeoff during which the airplane was exposed to continuing precipitation in below freezing temperatures. An objective determination of the amount of ice that could have fonned on the wings and empennage surfaces of the airplane after it was deiced requires analysis of numerous variables and assumptions. First, an estimation must be made of the length of time that the deicing fluid was effective. Although extensive research has been performed in ground deicing technology. the calculation of the effective holdover time of the deicing fluid is complicated by more than 30 variables that may influence the effectiveness of the deicing solution. Some of the more important variables after application include the influence of precipitation, deicing fluid thickness, strength, and temperature, aircraft skin and ambient temperature, wind (actual wind or apparent wind due to taxiing), residual moisture on airframe surfaces, and the conditioas of the ramp, taxiways, and runways.”?_ In addition, it has been shown that ice will not necessarily form at a uniform level across the wing, since ice accretion on a wing may Start earlier at certain locations than at others. Moreover, after the effective holdover time has been exceeded, the amount of precipitation accumulation on the airplane must be determined for the remaining time interva! before takeoff. Although the weather observatory at LaGuardia is about 3/4 of a mile from the gate area, the assumption was made that the rate of snowfall at the airplane location was consistent with that near the observatory. Other factors, such as aircraft skin temperature, shape and slope of the airplane surface, wind direction, and speed may also affect the accumulation of snow or ice on the airplane. The average amount of time calculated to deice/anti-ice an airplane was investigated.

[ANALYSIS] Conversely, a greater rate of precipitation accumulation would have had the effect of reducing the holdover time of flight 405. The Safety Board believes that given the numerous variables and complexity of the problem, the specific amount of ice that accumulated on the aerodynamic surfaces of the airplane during the taxi phase is indeterminable. However, the Safety Board also believes that some contamination occurred in the 35 minutes following the second deicing and that this accumulation led to the control difficulty shortly after rotation. 2.9 Flightcrew Performance - Takeoff Procedure and Stall Recovery The Safety Board views the evidence as conclusive that the primary factor in this accident was the reduced performance of the wing due to ice contamination. Therefore, the Safety Board evaluated the extent to which the decisions of, and procedures used by, the flightcrew could have contributed to the accident. After arriving at the USAir gate following the landing at LaGuardia, both the captain and the first officer departed the airplane for short pertods, and both of them were aware that the weather conditions were conducive to the accumulation of frozen precipitation on the wings. Upon retuming to the airplane, neither of them performed a walkaround inspection or took any special actions to check the condition of the wing leading edge and upper surface. However, the airplane was subsequently deiced and the wing condition was purportedly checked by ground personnel which obviated the need for the crew to depart the airplane a second time for an extemal inspection. That the captain requested a second deicing after about a 20-minute delay indicated his concem about the continuing exposure to precipitation; the request was prudent and in accordance with USAir guidance. Following the second deicing, the flightcrew was most likely satisfied that the airplane was free of adhering contamination. The flightcrew was not aware of the exact delay that they would encounter before takeoff and their decision to leave the gate was reasonable. After taxiing, when it became evident that they would be delayed for a prolonged period, conversations between the crew showed that they were aware of and probably concemed about the risk of reaccumulating frozen contamination on the wing. Their awareness of this risk should have been heightened by the need to use the windshield wipers intermittently in combination with the freezing outside air temperature. When it became apparent that the delay would exceed 20 minutes, USAir guidance prescribes a careful examination of the airplane's surfaces. The first officer stated after the accident, and passengers confirmed, that he had tumed on the wing inspection light to view the wing on several occasions.

[FINDINGS] At the time of the accident, procedures for opening emergency door exits were inaccurately and incompletely displayed on USAir's F-28 passenger safety briefing cards, but they did not contribute to the fatalities in the accident. The locations of the dike, pump house, and ILS localizer ground plane antenna were within current FAA guidelines; however, the locations did not meet [CAO Annex 14 criteria. The overall emergency response was effective and contributed to the survivability of the airplane's occupants; however, the response by the emergency medical services personnel was inadequately coordinated, and the ambulance response times to the hospitals were excessive. The difficulties that the air traffic controller experienced with the emergency telephone system did not hinder or delay the ARFF response. 3.2 Probable Cause The National Transportation Safety Board detennines that the prebabie causes of this accident were the failure of the airline industry and the Federal Aviation Administration to provide flightcrews with procedurcs, requirements, and criteria compatible with departure delays in conditions conducive to airframe icing and the decision by the flightcrew to take off without positive assurance that the airplane's wings were free of ice accumulation after 35 minutes of exposure to precipitation following deicing. The ice contamination on the wings resulted in an aerodynamic stall and loss of control after liftoff. Contributing to the cause of the accident were the inappropriate procedures used by, and inadequate coordination between, the flightcrew that led to a takeoff rotation at a lower than prescribed air speed. 78 4. RECOMMENDATIONS As a result of this investigation, the National Transportation Safety Board makes the following recominendations: --to the Federal Aviation Administration: If gate holds are required to limit deicing fluid holdover time, encourage air traffic control (ATC) to initiate the gate holds as soon as a deicing operation begins rather than after delays have exceeded 15 minutes, as in the current air traffic control definition of gate hold. (Class Il, Prionty Action) (A-93-19) Where deicing operations are conducted away from the departure runway, report taxi delays in conditions conducive to airframe icing in increments that are less than {{5 minutes to provide more realistic and useful reports to dispatchers and flightcrews. (Class H, Priority Action) (A-93-20) Require that flight crewmembers and appropriate ground personnel responsible for the inspection of transport-category airplanes for wing contamination receive specific periodic training that will illustrate what contamination looks like and feels like on a wing and the amount of contamination that is detectable under different light conditions. (Class IT, Priority Action) (A-93-21) Study the effects on performance of swept-wing turbojet airplanes when specific amounts of air speed are added to the computed roiation speed (delayed rotation) during takeoffs when wing contamination is possible. (Class HN, Priority Action) (A-93-22) Require Fokker to detemnine how takeoff performance and stall margin would be affected by using a lower initial target pitch attitude on F-28 and F-100 airplanes in the event that undetected upper wing ice contamination is present, and change the nommnal!

[ANALYSIS] Thus, it is probable that, during the transition to climb immediately after liftoff, the airplane reached an AOA beyond the stall AOA with significant loss of both lift and lateral control effectiveness. The abrupt roll that occurred during the takeoff of flight 405 is consistent with this analysis. The replication of events in the Fokker F-28 simulator confirmed that, with a contaminated wing, AOAs as high as 12 degrees, well into the stall regime, were reached even when the pilot initiated rotation at the proper speed to a target pitch attitude of 15 degrees ata rate of 3 degrees per second. The following is from a Fokker document”! on the effect of wing ice contamination on the F-28 wing: With frost roughness present on the wing upper surface the characteristic of slow stall progression towards the wing tip is lost and uncontrollable roll may develop at angle of incidence (attack) as low as 10 degrees...The drag of the clean wing is such that the aircraft is capable of climbing away at the required climb angle at V2 with one engine inoperative. In the case of a contaminated wing the drag may, however, be doubled due to a wing stall which occurs at an angle of incidence (attack) only slightly greater than that for 21 Fokker Report L-28-222 “Note on the Aircraft Characteristics as Affected by Frost, Ice or Freezing Rain Deposits on Wings,” dated December 16, 1969. stick shaker operation. Consequently, acceleration is lost even with ail engines operating at T.O. power. Most wings are designed so that the inboard sections will stall before the outboard sections. This design ensures that roll control can be maintained through use of the ailerons on the outboard wing sections. However, the variable distribution of ice particles and shorter chord length on the outboard sections of the wing usually create an irregular stall distribution across the wing. A premature stall of the outboard sections usually occurs first, with a resultant loss of lateral control. A significant nose-up pitching moment would also be expected in swept wing aircraft when the outboard wing sections stall. However, the sweep angle of the F-28 wing is only 16 degrees, and wind tunnel tests conducted by Fokker indicate that a nose-down pitching moment can occur following a contaminated wing stall. In any event, it was apparent from the evidence that after liftoff, the airplane could not transition to a positive climb angle during the 11 seconds that it was airbome before striking the dike. The maximum air speed recorded by the FOR during the 11 second flight was 134 knots, stick shaker activated at this time, and air speed then decreased and varied between 130 and 128 knots for the remainder of the flight. According to the Fokker simulation data, at this speed, the airplane should have been able to sustain a load factor of 1.5 G at the stick shaker threshold AOA which would still have provided about a 3-degree AOA stall margin.

[ANALYSIS] The conference produced suggestions for corrective actions that were taken before the 1992/1993 winter season and also offered possible long-term improvements to existing systems. The focus of the conference was on carrier-operated, turbine-powered airplanes with more than 30 passenger seats. From recommendations made by the working groups at the conference, on July 23, 1992, the FAA published a Notice of Proposed Rulemaking (NPRM) that would establish requirements for Part 121 certificate holders to develop an FAA-approved ground deicing/anti-icing program and to comply with that program any time such conditions as frost, ice, or snow could adhere to the aircraft's wings, control surfaces, propeilers, engine inlets, and other critical surfaces. If an air carrier does not want to have an icing program, they are given the option of performing a mandatory exterior icing check at least 5 minutes prior to takeoff for all flights, whether or not the airplanes were deiced/anti-iced prior to takeoff, when weather conditions are such that frost, ice, or snow could adhere to an airplane's critical surfaces. On September 29, 1992, an Interim Final Rule was published and became effective on November 1, 1992. In addition to the air carrier deicing programs required by rulemaking, the FAA is addressing the corollary issues relating to airport and air traffic control. Specifically, the actions being taken concem the reduction of the time that an airplane will be exposed to freezing conditions after having been deiced and the clearance for takeoff. This involves a reduction in ATC delays and, where practical, the implementation of offgate deicing facilities closer to the departure runways. 2.8.2 Reducing ATC Delays It is axiometic that the same weather conditions that prescribe the need to expedite an airplane's clearance for takeoff following a deicing operation are often the conditions most likely to lead to reduced aizport capacity and thus increased ATC delays. The FAA has acknowledged the need to address this 274 summary of the conference was published by the FAA Flight: Standards Service, Washington, DC 20591, in a document entitled “Report of the FAA Intematvonal Conference on Aisplane Ground Deicing.” 63 problem by reviewing ATC and airport procedures, such as gate hold and flow control. According to testimony, a departure delay is not initially reported by ATC until there is an actual delay of 15 minutes. The 15-minute delay does not include the addition of a “best case” (average) taxi time--which is inherent within the ATC system. For LaGuardia’s runway 13, the best case taxi time, from a gate to the takeoff end of the runway, is 8 minutes. Therefore, flightcrews preparing for departure on runway 13 can experience a ground time delay for as long as 23 minutes without an awareness that they will be delayed for takeoff.

[ANALYSIS] Although further study of aerodynamic stall margins and climb requirements is needed, decreasing the peak AOA during the takeoff maneuver would provide an enhanced level of safety for nonslatted airplanes taking off in icing conditions. There are far fewer nonslatted airplanes operating under {{4 CFR Part 121, but they have experienced almost all of the takeoff accidents attributed to wing upper surface ice contamination. Because of the critical nature of the takeoff maneuver in icing conditions, the Safety Board believes that the FAA, in conjunction with NASA, should establish a joint government/manufacturer task force to study methods to improve the AOA safety margin during the takeoff transition to initial climb. sore The FAA lias a téncécm about the cifects of advising pilots that nonslatted airplanes are more sensitive to wing ice contamination. It is believed that if nonsiatted wings are singled out, pilots will feel that a minute amount of ice is acceptable on slatted wing airplanes. The Safety Board agrees that operations with wing contamination should not be allowed or encouraged for any class of airplane. However, the icing accident record is worse for nonslatted airplanes, and differences in aerodynamic stall margins during the takeoff maneuver could explain the disparity in the accident record. Therefore, until research is completed, the Safety Board supports the requirement for a tactile or external close visual inspection of the wings of nonslatted airplanes immediately before takeoff when anti-icing holdover time has been exceeded. 70 2.8.5 New Technology for the Detection of Contamination The Safety Board has reiterated throughout this report the difficulties that flightcrews have in detecting contaminants on the wings of airplanes through visual inspection. To obviate the necessity of relying on visual perception or exterior tactile inspections to determine if the wings are clear of ice, snow, or frost, today's technology is being incorporated into equipment designed to detect contaminants on airplane surfaces and to present indications of unsafe conditions to ground personnel or flightcrews. Current concepts include electro-optical line-ofsight sensing extending into the infrared region or other techniques, such as measuring the changes in frequency and amplitude of vibrating piezo element diaphragms placed on the wing surface to detect contamination. The Safety Board believes that this technology is promising and will detect the presence of ice, snow, or frost on airplane surfaces. 2.9 Survivability 2.9.1 Passenger Safety Briefing Card The passenger safety briefing card did not show or describe how to operate the galley service door or the main boarding door in their emergency modes, which must be used when the doors fail to operate nonnally, as required by 14 CFR 121.571(4)(b)(1). The brieting card showed a plastic cover over the release handle for an overwing exit and an opening in the cover to permit its removal; in actuality, the opening is covered by thin plastic that must be broken before the cover can be removed. The_airplane's exit markings were in accordance with 14 CFR 25.811.

[ANALYSIS] Nevertheless, the Safety Board believes that at each airport the FAA identifies as likely to experience icing conditions regularly and with sufficient volumes of traffic, a deicing working group should be estuvlished, and maintained, and should meet regularly, especially before and during snow and ice seasons. These working groups should, at a minimum, include representatives from te ..at air carriers, fixed-base operators, FAA air traffic and airport safety and ce.’ .ication specialists, and airport management. The Safety Board believes that the FAA should require that each certificate holder, operating under Title 14 CFR Part 13°, whose airport ts determined likely to experience icing conditions regularly, establish and submit to the FAA for approval a deicing plan that includes, at a minimum, the membership of the airport deicing working group; the location(s), equipment, and procedures to be used for gate Jeicing and offgate deicing; description(s) of gate-hold parameters and procedures; and delineation of responsibilities for the deicing of airplanes at the gate or offgate, as applicable. SE A ~ s NE 29Fo¢ the purpose of this report. offgate deicing/anti-icing is defied as the elimination of ice/snow contamination on airplane fuselage. airfoil and engine surfaces, using Type Lor Type Il flutds, applied to airplane surfaces by fixed or mobile cguipment at an airport location away from the terminal/gate areas aad as close to the departure runway as is safely practicable, in order to reduce the elapsed time belween commencement of deicing/anti-icing and takeoff roll. 67 2.8.4 Pretakeoff Inspections of Airplane Wings The most positive assurance that an airplane is safe for takeoff in weather conditions conducive to the formation of frozen contaminants on the wing is a close inspection of the wing leading edge and upper surface immediately before takeoff. Federal Aviation Regulations require that the wing be clean; however, the investigation of past accidents has disclosed the difficulty involved with flightcrews determining whether wings are clean. The industry acknowledges that it is nzarly impossible to determine by observation whether a wing is wet or has a thin film of ice. While a very thin film of ice or frost will degrade the aerodynamic performance of any airplane, the Safety Board believes that the aerodynamic characteristics, as well as the accident record, indicate a need for special attention to be given to transport jet airplanes that do not have leading edge devices for lift enhancement during takeoff. The following is a general description of the effect of leading edge high lift devices, such as slats:*” An important (or predominant) limitation of lift to be obtained in wings, is flow separation from the leading edge.

[ANALYSIS] Other factors, such as aircraft skin temperature, shape and slope of the airplane surface, wind direction, and speed may also affect the accumulation of snow or ice on the airplane. The average amount of time calculated to deice/anti-ice an airplane was investigated. Based on past accident investigations, 12 minutes was the average time necessary to deice/anti-ice a large airplane using two deice/anti-ice trucks. It could take longer if there is a considerable accumulation of ice, the aliplane is large, such as a Boeing 747, and if only one truck is used. For smaller airplanes, deicing could take less than 12 minutes using two trucks. Aircraft exposure time must be calculated from the time that deicing begins rather than when it is completed. The FAA NPRM in the Federal Register of July 23, 1992, states that "Holdover time begins when aircraft ground deicing/antiicing commences and expires when the deicing/anti-icing fluid applied to the aircraft wings...loses its effectiveness.” 22FAA AC 20-117, “Hazards Following Ground Deicing and Ground Operations in Conditions Conducive to Aircraft Icing,” December 17, 1982. FAA AC 20-117 contains a simplified tonnula that gives a gross estimate of the length of time that a deicing fluid would be effective. However, according to the FAA, the formula tends to overestimate the holdover time at temperatures near freezing. To compensate for this limitation, the FAA introduced a cotrection factor of 0.5 into the forraula. Meteorological variables involved in the calculation include the precipitation rate and ambient air temperature. The precipitation rate of 0.09 inch water equivalent per nour was calculated from the weather observatory snowfall data. In addition, the lowest temperature of 29 degrees F, recorded by the Port Authority thermistor, was used for the calculations. The following table shows holdover times calculated using the formula given in AC 29-117 and the 0.5 correction factor for a range of Type I fluid thickness values using the precipitation rate and temperature values cited in the preceding paragraph. The thickness of the Type I fluid applied to the accident airplane is unknown. Type I Fluid Vhickness Holdover Tirne-- (mm) (Minutes) 0.020 4.55 0.025 5.67 0.030 6.83 0.035 7.96 0.040 9.10 0.045 10 24 0.050 11.37 0.055 12.5}} 0.060 13.65 0.065 14.79 0.070 15.92 0.075 17.06 0.080 18.20 NOTE: According to the FAA, a typical thickness for Type 1 fluid is 53 Temperatures greater than 29 degrees * would have increased the effective time of the deicing fluid.

[ANALYSIS] The Safety Board believes that a careful examination of the wings should involve some type of exterior inspection allowing for a close examination or tactile inspection if the holdover time has been exceeded. The Safety Board also believes that until more advanced technology is used to 7 detect ice accretion on wings, an additional deicing/anti-iciig is the only way to ensure that the wings are free of contamination prior to takec tl. Further, the Safety , Board recommends that the FAA require all operators to use (raining aids that will + illustrate to the flightcrew what contamination looks like and feels like on a wing, I and the amount of contamination that could be detected under different light conditions. Most pilots operating at LaGuardia during the time of the accident stated that they were checking other airplanes around their airplanes for snow/ice accumulation and were basing the decision to take off on the successful takeoffs of preceding airplanes exposed to the same weather. Yet, pilots have no means of knowing such critical details as the arrival/descent profile, ground time, gate exposure, deicing time, deicing fluid mix, and temperature of these airplanes. In ee short, the time history of other airplanes may be entirely different, and thus such comparisons are not valid. Moreover, the distances and lighting conditions make it Virtually impossible to detect the minute amounts of contamination that can adversely affect safe flight. ‘The Safety Board believes that the flightcrew of flight 405, as well as flightcrews of other airlines operating at the same time, did not have sufficient appreciation for the consequences that minute amounts of ice have on aircraft performance, notwithstanding the company training and literature on the subject, While the reference to a 9-degree AOA reached during takeoff in the GENERAL. section of the USAir F-28 Pilot's Handbook is accurate, the handbook tails to adequately stress its significance. The various forms of manufacturer literature, published since the manufacture of the F-28,26 identify liftoff at 8-degrees AOA, stick shaker at 12 degrees, and stall at 15-degrees AOA. Aerodynamic data from Fokker studies show that sandpaper-like contamination on the wing disrupts the normal stall progression toward the wing tip, and an uncontrollable wing roll may develop as low as 10-degrees AOA. The loss of control can occur betore stick shaker activation, and the pilot would not be aware that a stall is approaching until lateral control is lost. A note in the Pilot's Handbook implies that a smooth rotation would prevent pitchup and rolloff when contamination is present, and that smooth 26The first-production F-28 flew on 21 Miy 1968. continuous aileron, rudder, and elevator inputs would correct the problem. In tact, if there is contamination, a 3-degrees-per-second rotation rate can place the airplane into a stall regime at liftoff. The contamination-induced spanwise airflow negates the aileron effectiveness, and rudder input aggravates the stall. Although the clevator is still effective, the pilot has no altitude to trade for air speed.

[ANALYSIS] There is obviously a disagreement within the industry over the percentage degradation of lift due to upper wing surface contamination between slatted and nonslatted wings. However, there are no state-of-the-art wind tunnel results available to resolve this question. The Safety Board believes that the FAA, in conjunction with the National Aeronautics and Space Administration (NASA), should establish a wind tunnel and/or flight test program to study the aerodynamic degradation of both nonslatted and slatted airplane wings containing upper surface contamination. ie ee eS ee ee Eprom DC Flight Approach, "Wing Surface Roughness - Cause and Effect.” January, 1979, published by Douglas Aucraft Company. 2BLG. Ljungsuom, "Wind Tunnel Investigation of Simulated Hoar Frost on a Two-Dimensional Wing Section With and Without High Lift Devices,” FFA-AU-991, April, 1972. MBWingrips, page 6, December, 1989. Nonetheless, the critical factor in ice contamination is how close the takeoff maneuver gets the wing to its stall AOA. In this case, the fixed leading edge wing appirently has less margin of safety than the slatted wing, even if it is assumed that the percentage lift loss due to ice contamination is the same for both wings. During the takeoff maneuver, it takes longer to rotate the slatted airplane to a stall attitude so that the slatted airplane has time to climb and accelerate. Because airplanes with leading ecze slats normally stall at a higher AOA, the risk of an AOA overshoot into the stall region is lower than it is for a fixed leading edge airplane. The combination of more altitude, higher speed, and enhanced roll control increases the likelihood of a successful takeoff when the upper surface of a slatted wing is contaminated with a minimal amount of ice. Further, airworthiness requirements are based ona safe climb speed (V>) that is at least 20-percent above the stalling speed. Because the slatted wing creates lift over a broader range of AOA, a 20-percent margin in speed provides a slightly larger AOA margin before wing stall, typically a 1.5- to 2.0-degree greater margin between the AOA at V, and the stall AOA. Although further study of aerodynamic stall margins and climb requirements is needed, decreasing the peak AOA during the takeoff maneuver would provide an enhanced level of safety for nonslatted airplanes taking off in icing conditions. There are far fewer nonslatted airplanes operating under {{4 CFR Part 121, but they have experienced almost all of the takeoff accidents attributed to wing upper surface ice contamination.

[CONCLUSIONS] In both incidents, the following signs of the impending stall clearly existed : a) Continuous drag increase and correlative speed loss at constant engine power, b) Abnormal c) G-break, autopilot activity in roll prior to the disconnection, d) Stall warning. In both incidents, the flight crews readily recovered from the stall. None of them reported either abnormal or excessive aileron wheel forces during the recovery. The existence of icing conditions outside the JAR/FAR icing envelope was indirectly shown by computing the drag build-up from the DFDR data traces and by comparing it with the certification criteria envelope. The effects of the ice pollution were clearly shown to be beyond what had been taken into account in certification. Still, the stalls occurred at angles of attack consistent with the icing stall warning threshold. The “ice-induced aileron hinge moment reversal” phenomenon, which was discovered in the post-Roselawn accident investigation and testing, was not involved in the Air Mauritius and in the Ryanair incidents. Based upon the foregoing. neither of these incidents to ATR an icing scenario of the type of that was discovered during the p ost-Roselawn accident investigation 224 Since the failure of the flight crew to observe the minimum Np setting (86 %) in icing conditions was a common fact in both cases, at DGAC request, ATR rightly investigated the effects of the lower than required Np setting (77 % ) associated with severe icing conditions. ATR’s tests and research involved both theoretical studies and flight testing. The results were presented to the DGAC and to the BEA. ATR showed that the combined effects of Np setting at 77% and of severe icing conditions were likely to cause unusual ice accretions on the propeller blades, which in turn could generate an highly turbulent airflow over the wing airfoil. Since an increased level of turbulence is known to cause the deposit of a rough, thin layer of ice over the entire airfoil, especially in severe icing environments, this mechanism was believed to be the origin of the abnormal drag build-up observed prior to the Air Mauritius and RyanAir incidents and of the stall at or about the icing stall warning threshold. As a consequence, ATR took actions to re-emphasize the already existing limitations regarding the minimum Np setting in icing conditions. The aircraft check list was in particular amended for that purpose. The BEA, the DGAC, and ATR still believe that the mechanism described above could contribute to the alteration of aircraft performance in severe icing environments.

[ANALYSIS] Yet this same precipitation will flow aft on the wing and freeze creating a potentially dangerous situation. Crew vigilance must be used to detect the formation of ice as soon as possible. 188 The BEA has provided text of these critical documents in Section 1.4.1 and 1.4.2. Based upon these documents it is clear that American Eagle/Simmons’ policy mandated that flight crews exercise vigilance when operating in icing conditions and that flight crews avoid icing conditions when possible. There is a little doubt that the icing conditions encountered by Flight 4184 were at least “moderate” and possibly “severe”. In this regard, the Edwards flight tests demonstrated that operations in freezing drizzle conditions for a prolonged period of time, as was the case for Flight 4184, causes significant ice accretions to form on the frame of the aircraft’s windscreen, cockpit side windows, wiper blades, spinners, and ice detector probe. In this regard, Captain Jack Walters testified at the NTSB Public Hearing that Simmons’ flight crews are trained by American Eagle to look for these specific visual indicators to determine if the airplane is collecting ice. Proper monitoring of the outside air temperature, precipitation, and the ice accumulating on the aircraft should have informed the crew that they were operating in a freezing precipitation environment. These conditions were likely encountered by Flight 4184 and the flight crew should have requested a different altitude or holding pattern to avoid these icing conditions. The BEA considers that the crew did not observe the 14 CFR121.561 requirements relative to Pireps. 189 Finally, it is significant to note that the Airman’s Information Manual (AIM) specifically mandates that the crew of Flight 4184 “report icing conditions to ATC/FSS, and if operating IFR, request new routing or altitude if icing will be a hazard. ” Moreover had the crew of Flight 4184 provided ATC with a PIREP of their known icing conditions, it is reasonable to assume that on their request they would have promptly been issued a clearance and would have immediately exited the area, thus avoiding the accident. 2.2.3.5. THE DFDR AND CVR DISTRACTED WHILE DATA SHOW THAT THE FLIGHT CREW WAS MANAGING THE AIRCRAFT'S DE-ICING AND ANTI-ICING EQUIPMENT IN ICING CONDITIONS The BEA analysis of Flight 4184 crew operation of aircraft Systems is made in reference to criteria established by the NTSB, presented in section 1.11 of this Document, as Primary Error No 7. There is no ATR's AFM issue about the definition of icing conditions. provides specific procedures in respect to operation of the ATR72’s anti-icing system (Level II) and de-icing system (Level III). ATR's AFM, Section 3.04 Procedure for operation in atmospheric icing conditions, provides that Level II anti-icing systems, which consist of Propeller 1 and 2, Horn 1 and 2, Engine 1 and 2, Side Window and NP 86% minimum, are to be activated when icing conditions exist.

[CONCLUSIONS] At that time the conclusion was that unusual ice accretion patterns may have been caused by the speculated aircraft’s sustained flight in freezing rain conditions, and it was concluded that such conditions could have been the origin of an aileron hinge moment modification which occurred about at the stall threshold when the autopilot disconnected. However, the absence of documented freezing rain ice shapes and of any industry standards for such ice accretions gave no basis to support or test such speculation. Several other factors limited any further analysis of this matter by ATR, or by any other party involved in the investigation, including the NTSB. 220 a) The worldwide industry belief that freezing rain conditions, which are beyond the certification envelope of all aircraft, were rare occurrences, and that it would be impractical to protect aircraft against their effects. Further, it was believed that such conditions were generally “predictable, recognizable and avoidable” [AC 20.117]. b) The absence of certification criteria to cover the consequences of inadvertent encounters and the absence of documented effects in terms of ice accretion patterns. c) The fact that the event occurred at about the stall threshold, which was further addressed by the vortex generators modification. d) The fact that the crew had failed to activate the airframe de-icing equipment at the time, which fact was revealed by the NTSB to the other investigating parties after the manufacturer’s analysis was published, was an aggravating factor in the incident. Application of the normal and required de-icing procedures for flight in icing conditions may well have prevented the incident. e) The fact that the crew had not reported abnormal or excessive wheel efforts during the recovery and that the aileron effectiveness had apparently remained unaffected. In the frame of the post-Roselawn accident investigation, the BEA made further inquiries about potential similarities between the ATR 42 Mosinee incident and the ATR 72 Roselawn accident and considered whether additional efforts might have allowed the investigating parties at that time to anticipate the Roselawn icing scenario. 221 Several factors made the ice accretion patterns involved in the Mosinee incident, significantly different from those that most probably developed in the Roselawn case, the first of which being the fact that aircraft S/N 91 accreted ice with the airframe de-icing boots OFF. Also, aircraft S/N 91 accreted ice in the flaps 0° configuration ; droplets sizes associated with the prevailing freezing rain conditions were probably higher than those involved in Roselawn ; the exposure time was not longer than 10 minutes. These differences resulted in ice accretion patterns with both large spanwise and chordwise extents on the wing airfoil and with limited protruding ridge height. Although such shapes cannot be accurately characterized, the BEA believes that their nature may not be very different from one of the Edwards tanker test cases, with the wing de-icer boots inoperative (test n°23) exhibited.

[CONCLUSIONS] However, the stall nevertheless occurred at the icing stall warning threshold and a massive drag build up and the correlative airspeed loss should have triggered the crew’s attention. Such severe drag increases were felt to be alwavs associated with these unusual ice accretions, as all incidents had indicated, including this one. Also, recovery actions were readily accomplished by the flight crew, and aileron effectiveness and control wheel forces were not reported to be abnormal. 228 Nevertheless, because the ATR-72 had no in-service history of any such roll control icing related events and was certified using different and modified icing codes from those used for the certification of the ATR-42, the DGAC required that ATR re-visit the determination of ice accretions within the Appendix C envelope, under the modified codes, for the ATR-42. This research was on-going at the time of the Roselawn accident. 229 2.3.5. CONCLUSION Until the post-Roselawn tests and research, the freezing drizzle/freezing rain conditions were not perceived by the worldwide industry as a major threat to the safe operation of regional airline aircraft. These conditions were, and still are, omitted from the certification criteria. Although such conditions were generally understood to be hazardous and to be avoided, there was no absolute prohibition to fly into such conditions, based upon the assumptions that they were rare occurrences which could be recognized and avoided and that properly certificated aircraft would safely cope with short inadvertent encounters. Neither regulatory environment nor the available means of experimental research did encourage the Western manufacturers and Airworthiness Authorities to focus on the characteristics of such conditions and on their potential effects on the aircraft performance and controllability, but rather, to re-emphasize good operational practices to avoid such conditions as ATR has emphasized, and in particular, re-emphasized following the Mosinee incident. The BEA notes that among of the ATR 42 incidents, which all occurred in the clean, flaps 0°, configuration, and which all involved failure to follow icing conditions procedures none, exhibited the unique characteristics involved in the Roselawn accident, namely an outer wing airfoil flow separation at an AOA well below the icing stall warning threshold, without any prior noticeable drag build-up and without any significant asymmetrical lift loss. 230 Aileron hinge moment modifications could only be noted in two of these incidents - it is still doubtful that they existed in the Newark incident - and their contribution was perceived bv all investigating parties as a marginal characteristic of what was substantially concluded as an asymmetrical stall in severe icing conditions. There was no evidence that such modifications of the aileron hinge moment could become a predominant factor in different circumstances, since they had not initiated roll excursions or interfered with crew recovery actions in any of the incidents. It is furthermore today understood that the prior ATR 42 incidents and the ATR 72 Roselawn accident involved different mechanisms and amplitudes of airflow separations.

[CONCLUSIONS] Based upon the foregoing, the BEA concludes that aircraft S/N 401 was fully recoverable. 216 2.2.5. CREW RESPONSE TO ROLL UPSET Aircraft recovery following SLD induced roll upsets have been flight tested with the Edwards testing simulated ice shapes glued on the wings : aircraft recovery was repeatedly shown to be physically possible, with one pilot alone using his yoke only. Such a recovery requires a firm pilot’s action to overcome the jerky forces which otherwise would drive the roll control wheel in the direction of the upset. Time reaction is also critical, as the roll upset, if not counteracted in the first few seconds, could develop a high rate of roll which would aggravate the pilot disorientation. 217 2.3. PREVIOUS ATR ICING INCIDENTS AND ADEQUACY OF DGAC/ATR ACTIONS The following analyses of previous ATR-42/72 incidents incorporates the results of all tests and research conducted after the Roselawn accident Therefore, these analyses may also review some of the assumptions formulated during the previous analyses of these events. 2.3.1. MOSINEE INCIDENT The Mosinee incident was the first of five events analyzed by the NTSB, experienced by an ATR-42 aircraft in icing conditions outside the icing certification envelope. These conditions were clearly shown to be freezing rain, associated with a temperature inversion phenomenon. This incident involved an auto-pilot disconnection, at an angle of attack very close to the icing stall warning threshold, with evidence from the DFDR data traces of a rolling moment induced by an asymmetrical lift loss and with evidence of an aileron self deflection. The recovery was readily accomplished by the flight crew. The investigation later revealed that the flight crew had not activated the airframe de-icing equipment prior to the incident, while the aircraft was accreting ice. ATR’s initial response to this first incident was to immediately reemphasize to all operators the hazards associated with flight operations in freezing rain using the FAA’s Advisory Circular 20.117 material. ATR subsequently proposed a design modification to the ATR 42 (Vortex Generators) and also and to proposed changes to the AFM/FCOM to incorporate procedures applicable to inadvertent encounters with such conditions. 218 These proposed changes were submitted to the DGAC, which in turn submitted the proposed manual changes to other Airworthiness Authorities, including the FAA. The FAA, as stated in its subsequent NPRM (Notice of Proposed Rulemaking) regarding the introduction of the vortex generators modification, declared that as a matter of policy, hardware changes were preferred to procedural changes.

[CONCLUSIONS] After the Mosinee incidents, ATR proposed to the FAA, through the DGAC, a revision to the ATR-42 FCOM and AFM which contained information on the effects of freezing rain conditions on aircraft stability and control characteristics and on the autopilot and set forth related operational procedures to be used when an aircraft inadvertently encounters such prohibited conditions. This proposal was not accepted by the FAA. 19. ATR provided Simmons and other operators with the identical information, applied to both the ATR-42 and ATR-72 aircraft, concerning the effects of freezing rain (understood by Simmons to include “freezing precipitation” in the AOM). 261 20. ATR provided specific warnings to Simmons and other operators, for their pilots, about the adverse characteristics of freezing rain and about roll events which could occur in such conditions and gave specific guidance for recovery from such events and, in addition, developed aircraft modifications seeking to reduce the possibility of such events occurring. 21. Simmons company policy had already provided ample instructions to the Flight Crews regarding the icing threat and the basic rules of behaviour to face such a situation. 22. The failure of Flight 4184’s flight crew to follow these company policies and manual provisions and exit the known icing conditions led directly to this accident. 23. Despite the lack of anticipation by the NTSB, BEA, ATR, FAA and DGAC, prior to the accident, of the mechanism of the ice-induced aileron hinge moment reversal, Simmons/AMR Eagle and its flight crews had been warned that, under icing conditions outside those specified in 14 CFR Part 25. Appendix C the ATR 42/72 aircraft performance and controllability might be affected in such a way that auto-pilot self-disconnect and subsequent roll excursions could occur; that roll efficiency would nevertheless be maintained; that recovery could be achieved by making firm aileron inputs to counter the roll excursions and by applying basic stall recovery techniques. These were appropriate and adequate instructions to flight crews based on what was known from prior incidents. 262 24. ATR adopted appropriate and adequate changes to its flight crew training program and simulator data known from prior icing incidents. training package based on what was 25. Chicago ARTCC controllers were aware that light to moderate icing conditions were forecast for the area of LUCIT intersection at the time Flight 4184 was released from its ground hold. 26. Chicago ARTCC controllers had received PIREPs reporting icing conditions on the day of the accident and had their supervisor at the beginning of their shift icing conditions and because “Icing Kills”. been specifically briefed by that they must be aware of 27. Chicago ARTCC controllers were aware that the weather conditions were deteriorating throughout the Chicago area before and during the time Flight 4184 was enroute from Indianapolis to Chicago. Therefore they could not have ignored the specific weather conditions at the Lucit holding pattern, at Flight Level 100. 28.

[ANALYSIS] These systems must be activated prior to the activation of the Level III De-icing System which consists of the airframe de-icing system (boots) and which are used only when ice starts to actually accrete on the aircraft. 190 The DFDR data indicates that at 1516:32, the airframe de-icing system was activated. This means that Flight 4184 was probably accreting ice. At 1524:30, the airframe de-icing system was turned off. At 1524:50, twenty seconds later, the flight crew selected NP 7 7 % whilst they still were in icing conditions. According to ATR's AFM and AOM procedures, NP 77% is not to be selected in icing conditions. NP 8 6 % at least, on the other hand, must be utilized when the Level II anti-icing systems are activated. Based upon this information, the BEA suggests that two separate hypotheses are possible in respect to the crew’s actions: (1) the remaining components of the Level 11 anti-icing system were de-activated (Engine 1 and 2, Propeller 1 and 2, Horn 1 and 2, and Side Window) at or about the same time the crew selected NP 77%; or (2) the flight crew left the remaining components of the anti-icing system ON. The BEA has that this had At 1528:00, “single tone investigated and analyzed prior icing incidents and has found occurred in the past. the CVR transcript starts. At 1533:56, similar to a caution alert chime. ” The the CVR recorded a DFDR data and the meteorological conditions, information set forth in the BEA’s meteorological study show that between 1523 and 1534, Flight 4184 was intermittently operating in liquid precipitation with a SAT between -2.5 and -4,0 degrees C. During this time period, the flight crew maintained NP 77% and the airframe de-icing was deactivated. This lack of action by the flight crew was in violation of the ATR-72 AFM/AOM. In this regard, the Simmons Airlines Winter Operations Handout also specifically provides : 191 Crews are cautioned to remain alert to these conditions and frequently check Static Air Temperature (SAT) indications during cruise and descent. If SAT indications reach a value of 5 degrees C or less, good operating practice would dictate that icing equipment be operated accordingly.

[ANALYSIS] The excessive duration of the flight in such conditions, with no recorded comments (as shown by the CVR transcript) on the severity of the icing, nor any upon the procedures to be applied in the conditions, seems incomprehensible on the part of the flightcrew. Another major element is the domain of aircraft certification in icing conditions. The reference is appendix C of JAR - JAR 25 regulation, which sets the certification limits. This regulation does not consider the existence of supercooled droplets or drops having a diameter over 40µm (continuous maximum atmospheric icing conditions) with a liquid water content over 0.8 g/m3 in the cloud layer nor the case of freezing drizzle or freezing rain. 174 Thus the BEA's study points up the following five findings : 1. According to the content of the flight release, the crew was aware of the existence of light to moderate icing on the Indianapolis - Chicago route at the levels at which they were flying. 2. In an available AIRMET, valid before and for the flight, rainfall was forecast at the altitude of Flight 4184 with negative air temperatures. 3. Precipitation was detectable on the airborne radar on WX position. 4. The flight in the holding pattern lasted over 30 minutes in a cloudy atmosphere with liquid precipitation and at a SAT varying betwwen -2 and - 4°C. This was in complete contradiction with the limits specified in the certification and operational procedures. 5. Procedures relative to flights in icing conditions, specifically those related to the surveillance of environment, static temperature, ice indicators, and detectors, as well as some visual cues, were not respected by the flightcrew. In addition, standard procedures relating to propellor speed adjustment and anti-icing and de-icing system activation in icing conditions were not properly applied. In conclusion, overall crew vigilance and awareness did not correspond to the basic rules to be applied on such a flight, occurring in icing conditions conducive to ice accretion. 175 2.2.1. HOLDING TECHNIQUE The holding pattern is flown with Auto-Pilot in the Altitude-Hold Mode. Under these conditions, the airspeed must be maintained by manual adjustment of the engine torque. It is observed that throughout the entire holding patterns the number of these adjustments is very limited. As a consequence, airspeed variations of more than 10 knots are noticed during each holding turn, leading the airspeed to decay marginally below the minimum authorized speed (Vm HBO-icing) which was computed at 165 Kt for this holding. Utilization of higher holding speeds, which could have been authorized by the Air Traffic Controller, would have minimized the crew’s feeling related to the aircraft “wallowing in the air”, even during the phases where airspeed was reduced as the result of their limited power adjustments. This higher holding speed would have precluded the flight crew’s ad hoc decision to use a different flap setting than the one provided for in the aircraft manuals and which was initially selected by the crew. This would have increased the safety margin with the minimum authorized speed while eliminating the risk of inadvertently reaching the maximum authorized speed limit.

[CONCLUSIONS] The BEA discusses its further comments regarding this issue below. Despite the lack of identification by the NTSB, BEA, ATR, FAA, and DGAC, prior to the Roselawn accident of the freezing drizzle induced “aileron hinge moment reversal” phenomenon, the documents discussed in Sections 1.4.1 and 1,4.2 above clearly show that American Eagle/Simmons passed on to its flight crews these ATR warnings that, under icing conditions outside those specified in 14 CFR Part 25, Appendix C, the ATR-42/72 aircraft performance and control forces may be affected in such a way that autopilot self-disconnect and subsequent roll excursions could occur; that roll efficiency would nevertheless be maintained; and that recovery could be readily achieved by making firm aileron inputs to counter the roll excursions, and by applying basic stall recovery techniques. 255 Simmons own (ref. Appendix “restatement of company operations policies” 2) further provided : a) “Large droplets of freezing rain impact much larger areas of aircraft components and will in time exceed the capability of most ice protection equipment”; b) “Flight in freezing rain should be avoided where practical”; c) “If icing or adverse weather is experienced, make a PIREP . . “; d) “Freezing rain may form ice on an aircraft that is near the freezing level”; e) “If freezing rain is encountered, you should exit the condition immediately. This diversion should consist of a turn towards better conditions and/or climb to a warmer altitude”; f) “Freezing rain and clear ice can be very difficult to recognize on an aircraft, therefore it is strongly recommended when operating in conditions favorable to this type of icing that an extra vigilance be maintained;” g) “However, our aircraft are not to be operated in known freezing rain or severe ice. If these conditions these conditions immediately. ” are experienced, the procedure is to exit Flight 4184’s flight crew violated these “company operations policies” by not avoiding freezing precipitation conditions; by not making a PIREP; by not exiting the freezing precipitation conditions immediately, and most importantly, by not exercising crew vigilance in such conditions.

[ANALYSIS] This lack of action by the flight crew was in violation of the ATR-72 AFM/AOM. In this regard, the Simmons Airlines Winter Operations Handout also specifically provides : 191 Crews are cautioned to remain alert to these conditions and frequently check Static Air Temperature (SAT) indications during cruise and descent. If SAT indications reach a value of 5 degrees C or less, good operating practice would dictate that icing equipment be operated accordingly. Further, while the aircraft was flying the second hold, from 1532:30 to 1534:30, a large decrease northern right turn in the of 14 kts was recorded on the DFDR. The analysis of this speed reduction indicates that it was caused by the following : turn technique initially conducted at constant torque with a high bank angle representing a large contribution to the speed decay; fluctuating winds with a magnitude of up to 40 kts from the south southwest (210 degrees); ice accretion resulting from icing conditions with freezing precipitation confirmed by the BEA study. Although it was impossible to accurately evaluate the different individual contributions to the large speed decrease, there is no question that part of this speed decrease was attributable to the ice accretion. The NTSB’s own analysis indicates a first small drag increase 24 minutes before the roll upset, at 15.33. Thus, the 1533:56 caution alert chime might have corresponded to the aircraft’s ice detector system (AAS), which would have responded within 30 seconds after the first ice accretion began. This view is supported by the fact that during the Edwards AFB flight tests, the AAS system activated the aural icing warning within 30 seconds of encountering the artificial freezing precipitation conditions. Significantly. the 1533:56 caution alert chime was never acknowledged by the flight crew according to both the CVR and the DFDR data. 193 At 1541:07, a “single tone similar to caution alert chime” sounds. The DFDR data indicates that the flight crew then activated the airframe de-icing system, Two seconds later at 1541:09, the flight crew increased NP from NP 77% to NP 8 6 % . This DFDR data is confirmed by the increased noise which can be heard on the CVR tape. AT 1542:02, the CVR transcript records “8 clicks” which could have corresponded with the activation of the following anti-icing systems: pushbuttons for Engine 1 and 2, Propeller 1 and 2, Horn 1 and 2, Side Window, and the engine continuous relight knob. However, during the BEA’s investigation, the BEA had the opportunity to participate in two test flights during which the BEA listened and recorded on the CVRs various sounds generated in the cockpit. No clicks were audible on the CVR. Moreover the activation of two push-buttons of the airframe de-icing system is not audible on Flight 4184’s CVR.

[CONCLUSIONS] These differences resulted in ice accretion patterns with both large spanwise and chordwise extents on the wing airfoil and with limited protruding ridge height. Although such shapes cannot be accurately characterized, the BEA believes that their nature may not be very different from one of the Edwards tanker test cases, with the wing de-icer boots inoperative (test n°23) exhibited. In this respect : a) both the Edwards tanker test (N°23, Flap 15 degrees) and the subsequent corresponding flight test in Toulouse with artificial ice shapes directly derived from the observations made at Edwards, show handling qualifies effects consistent with the Mosinee DFDR data, in that the roll control is not affected prior to an AOA very close to the icing stall warning threshold, b) all available wind tunnel and flight tests data indicate that unusual ice accretion patterns with a large spanwise coverage would noticeably increase the drag, prior to any lateral control alteration. Such was the case in the Mosinee incident. 222 c) all available wind tunnel and flight tests data indicate that unusual ice accretion patterns of that same type would generate high lift losses, of a genuine asymmetrical nature. Such was the initiating factor of the roll departure in the ATR 42 Mosinee incident. d) both the Edwards tanker test (n°23) and the subsequent flight test in Toulouse with artificial ice shapes directly derived from the observations made at Edwards, show some degree of aileron hinge moment shift after the initiation of the roll motion due to the asymmetrical lift loss. Such was the case in the Mosinee incident. The BEA therefore concludes that the ice accretions patterns, that the mecanism of the airflow disturbance generated by these ice shapes, that the resulting handling effects, involved in the ATR 42 Mosinee incident were different from the ice shapes, airflow separation and hinge moment reversal revealed by the post-Roselawn investigation. As the consequence, should the investigating parties in the Mosinee incident have decided to conduct further testing and should have testing means been available - which was not and is still not the case - the BEA believes that the simulation of the Mosinee incident - flaps 0° configuration, de-icer boots OFF, freezing rain droplets - might have reproduced the characteristics of this incident but would not have allowed the anticipation of the Roselawn icing scenario. 223 2.3.2. AIR MAURITIUS AND RYAN AIR INCIDENTS The analysis of the next two ATR-42 incidents showed that the roll excursions were primarily caused by asymmetrical lift loss. No alteration of the aileron hinge moment was evident from the DFDR data traces, nor can such evidence be seen todav in re-visiting the analyses of these traces. In both incidents, the following signs of the impending stall clearly existed : a) Continuous drag increase and correlative speed loss at constant engine power, b) Abnormal c) G-break, autopilot activity in roll prior to the disconnection, d) Stall warning. In both incidents, the flight crews readily recovered from the stall.

[CONCLUSIONS] The release of Flight 4184 under these conditions was contrary to the policy established in FAA Order 7110.65, Air Traffic Control, to reduce congestion in the air traffic system and to limit the duration of airborne holding. 259 9. American Eagle/Simmons’ policy precluded the distribution of AIRMET Zulu Update 3 for icing and freezing level in the Flight Release for Flight 4184. This AIRMET was applicable to Flight 4184’s route of flight from Indianapolis to Chicago, and stated that “light occasional moderate rime icing in cloud and in precipitation” could be expected. This AIRMET also provided information regarding the freezing level along Flight 4184’s route of flight. 10. AMR Eagle/Simmons was adequately warned by ATR prior to the accident about the dangers of operating in freezing precipitation and understood the need to avoid such conditions. 11. AMR Eagle/Simmons, in turn, warned its flight crews prior to the accident about the dangers of operating in icing conditions, including freezing precipitation, and instructed its flight crews to avoid such conditions. 12. The flight crew of Flight 4184 had been expressly warned about the dangers of freezing precipitation and the necessity of crew vigilance. 13. Flight 4184’s flight crew knew they were operating in icing conditions. 14. Proper monitoring of the outside air temperature, clouds, precipitation, and the ice accumulating on the aircraft by the crew of Flight 4184 would have informed them that they might be operating in a freezing precipitation environment. 260 15. Despite these warnings and instructions, and having entered known icing conditions, the flight crew of Flight 4184 had absolutely no discussions regarding: the nature and extent of the icing conditions they were encountering; the outside meteorological conditions; the need to request a clearance to an alternative altitude or route to remain clear of the known icing conditions; the operation of the aircraft’s de-icing and antiicing equipment. 16. Flight 4184’s flight crew had ample opportunity to ask the ATC for a clearance to exit the icing conditions. 17. AMR Eagle/Simmons’ company policies require that flight crews stay out of icing conditions when possible. 18. After the Mosinee incidents, ATR proposed to the FAA, through the DGAC, a revision to the ATR-42 FCOM and AFM which contained information on the effects of freezing rain conditions on aircraft stability and control characteristics and on the autopilot and set forth related operational procedures to be used when an aircraft inadvertently encounters such prohibited conditions. This proposal was not accepted by the FAA. 19.

[CONCLUSIONS] This directly contributed to the accident. 37. The Captain’s lack of assertiveness and complete failure to integrate himself into the required flight activities left the entire operation of the aircraft to the First Officer. 38. AMR Eagle/Simmons’ ATR42/72 Airplane Operating Manual (AOM) provides only for holding with the aircraft configured in the flap zero degree configuration. Flight 4184’s flight crew’s unauthorized use of the flap 15 configuration while holding at 175 knots in icing conditions created the critical ice ridge beyond the de-icing boots which ultimately led to the roll upset, and thereby directly contributed to the accident. 39. Post-accident flight tests at Edwards Air Force Base and in France confirmed that Flight 4184 was recoverable after the initial roll upset. 265 3.2. PROBABLE CAUSE This accident was caused by a combination of factors, as reflected in the following BEA-proposed Probable Cause Statement : The Probable Cause of this accident is the loss of control of the aircraft by the flight crew, caused by the accretion of a ridge of ice aft of the de-icing boots, upstream of the ailerons, due to a prolonged operation of Flight 4184 in a freezing drizzle environment, well beyond the aircraft’s certification envelope, close to VFE, and utilizing a 15 degree flap holding configuration not provided for by the Aircraft Operating Manuals, which led to a sudden roll upset following an unexpected Aileron Hinge Moment Reversal when the crew retracted the flaps during the descent. The contributing factors to this highly unusual chain of events are : 1. The failure of the flight crew to comply with basic procedures, to exercise proper situational awareness, cockpit resource management, and sterile cockpit procedures, in a known icing environment, which prevented them from exiting these conditions prior to the ice-induced roll event, and their lack of appropriate control inputs to recover the aircraft when the event occurred ; 2. The insufficient recognition, by Airworthiness Authorities and the aviation industry worldwide, of freezing drizzle characteristics and their potential effect on aircraft performance and controllability ; 3. The failure of Western Airworthiness Authorities to ensure that aircraft icing certification conditions adequately account for the hazards that can result from flight in conditions outside 14 CFR Part 25, Appendix C, and to adequately account for such hazards in their published aircraft icing information; 4. The lack of anticipation by the Manufacturer as well as by Airworthiness and Investigative Authorities in Europe and in the USA, prior to the post accident Edwards AFB testing program, that the ice-induced phenomenon could occur. 5. The ATC’s improper Flight 4184. Aileron Hinge moment reversal release, control, and monitoring of 267 4. RECOMMENDATIONS The BEA notes with interest the disparity between the broad scope of the recommendations which the NTSB makes as a result of this accident and the selective focus of the NTSB’s statements of its findings and proposed Probable Cause of this accident.

[CONCLUSIONS] Accordingly, none of the investigating parties, including the NTSB, BEA, DGAC and ATR, could identify the exact contribution, if any, of an ice-induced pollution of the airframe in the Newark incident. None of the same parties which investigated this event had any indication that an aileron hinge moment modification could be a significant factor as it was in the Roselawn accident. Following the Roselawn accident, the BEA and NTSB reviewed the Newark DFDR. Because the DFDR readout disclosed that a high level of turbulence was involved throughout the incident, and would by itself explain the aircraft behavior, it cannot be determined whether the ’’ice-induced aileron hinge moment reversal” phenomenon which was discovered for the first time in the post-Roselawn accident investigation and testing was involved at all in the Newark incident. 227 2.3.4. BURLINGTON INCIDENT This incident occurred early 1994 and its DFDR data was reviewed by ATR and the BEA. The aircraft experienced, prior to the autopilot disconnection, a continuous speed decrease of about 45 kts, without any pilot corrective action. The airspeed reached prior to the upset was below the minimum prescribed speed for icing conditions. In addition: a) a g-break was apparent before the A/P disconnection, b) the autopilot was disconnected by the stall warning, c) the aileron briefly self-deflected after the stall commenced. ATR identified in its analysis a momentary aileron hinge moment modification. However, the predominant factor was clearly the asymmetrical lift loss in the stall which induced the roll motion. The momentary modification in the aileron hinge moment which occurred after the stall commenced had no effect on this incident. d) The Np setting was 86%, in accordance with the published procedures. but the airspeed was below the specified minimums. This incident was considered by the BEA, DGAC, and by ATR as an indication that unidentified ice accretion patterns, other than that caused by a turbulent airflow behind an improperly de-iced propeller might alter the aircraft performance and controllability. However, the stall nevertheless occurred at the icing stall warning threshold and a massive drag build up and the correlative airspeed loss should have triggered the crew’s attention. Such severe drag increases were felt to be alwavs associated with these unusual ice accretions, as all incidents had indicated, including this one.

[CONCLUSIONS] After Roselawn, EDWARDS tests revealed the particular and very specific type of ice accretion resulting from prolonged exposure to SLD with 15°flap. These holding conditions, never provided for in the Aircraft Operating Manuals led to a negative AOA which generated quite unusual ice shapes, in that the accretion concentrates on the upper wing aft of the deicing boots and with limited coverage of the wing lower surface. - The severe anomaly in roll which was discovered at EDWARDS results from the following unique sequence : prolonged ice accretion phase in SLD conditions with Flap 15° configuration and stall demonstration performed at the Flap 0° configuration. - Given the technology of unpowered flight control systems, all Commuterclass turboprop are affected by the same type of roll control problem, when submitted to the same SLD environment and same configuration changes. 248 - After full understanding of such a complex icing process, the ATR deicing system was modified with extended overwing boots which were tested successfully at EDWARDS in SLD environment. - Associated with AFM procedural changes (visual cues, flaps utilization), they provide the ATR fleet with a demonstrated level of safety in case of inadvertent encounter with SLD conditions, which is beyond the current icing certification standard. This physical modification associated with these procedural changes have been recognized as an acceptable means of compliance, and therefore a terminating action to the ADs respectively issued by DGAC and FAA. Recent industry tests and research conducted after, and as a result of the Roselawn accident, have provided valuable information on the potential effects of unusual ice accretions in the SLD environment. In light of this new information, the BEA understands that certification criteria will be changed to better address these conditions in line with the recommendations of the FAA/DGAC Special Certification Review (SCR) report. Changes in the regulatory standards are therefore being prepared in both France and the US to : - identify the physical characteristics associated with large supercooled droplets outside of Appendix C conditions. - establish criteria for acceptable aircraft behaviour in the presence of accretions resulting from these extreme conditions, as well as the same associated means of compliance. 249 2.7. ATR DISSEMINATION OF 2.7.1. THE NTSB REPORT MISREPRESENTS FACTS AND ATR KNOWLEDGE The Report’s probable cause finding (and the associated analysis and findings) that ATR failed to completely disclose to operators “adequate information concerning previously known effects of freezing drizzle and freezing rain conditions on the stability and control characteristics, autopilot and related operational procedures when the ATR 72 is operated in such conditions” is supported by the NTSB’S record of investigation and is wrong.

[ANALYSIS] The Captain was subsequently absent from flight deck for 5:25 minutes. During this time he engaged in a “non-essential conversation” which had no bearing on the safe operation of the aircraft. The Captain returned to the cockpit at 1554:13, approximately two minutes later. Approximately one and one half minutes after the Captain returned to the flight deck, the Co- Pilot stated “we still got ice” at 15:42. The Captain did not acknowledge this comment. Based upon the foregoing, it is clear that the Captain was not exercising proper situational awareness or proper vigilance in respect to monitoring the icing conditions. By leaving the cockpit, the Captain also deprived himself of any opportunity to monitor and request a clearance to deviate from the icing conditions. At no time while the Captain was at the back of the aircraft on the intercom with the Co-Pilot or, when he returned to the cockpit, did the Captain inquire about the icing conditions. Further, there is no indication that the Captain observed the aircraft’s propeller spinners or any other visible part of the airframe for ice accretion while he was walking back and forth through the aircraft cabin. The Captain’s lack of vigilance in this regard was directly contrary to American Eagle/Simmons’ policies discussed above. 199 It is very significant that there is no evidence on the CVR transcript that the flight crew discussed the operation of the aircraft’s deicing and anti-icing equipment or, that they monitored or discussed the type of ice accumulation or ice accretion rate. Further, there is no evidence that the crew notified ATC that they had encountered icing conditions or, considered giving ATC the Pirep required by Simmon’s policies or that they discussed any alternative altitude, holding pattern or route to exit the icing conditions. In this regard, the FAA Principle Operations Inspector (POI) for Simmons Airlines testified at the NTSB Public Hearing and responded to questions regarding various flight related functions perform by the crew of Flight 4184. The POI stated that given the environment in which Flight 4184 was operating in, “. . . I think I would expected more exchange [verbal communication] between the First Officer and the Captain about the amount - that the ice was there . . .” Finally, by leaving the cockpit, the Captain completely lost what little situational awareness he had regarding the operation of the flight.